Source code for biogeme.expressions

""" Defines the various arithmetic expressions accepted by Biogeme.

:author: Michel Bierlaire

:date: Tue Mar 26 16:47:49 2019
"""

import logging
from itertools import chain
import numpy as np
import cythonbiogeme.cythonbiogeme as ee
import biogeme.exceptions as excep
from biogeme.idmanager import IdManager
from biogeme.elementary_expressions import TypeOfElementaryExpression
from biogeme.controller import CentralController

logger = logging.getLogger(__name__)


[docs]def isNumeric(obj): """Checks if an object is numeric :param obj: obj to be checked :type obj: Object """ return isinstance(obj, (int, float, bool))
[docs]def process_numeric(expression): """Transforms a numeric value into an Expression object :param expression: expression to process :type expression: Expression or numeric :raise BiogemeError: if expression is not of type expression """ if isinstance(expression, (int, float, bool)): return Numeric(expression) if not isinstance(expression, Expression): raise excep.BiogemeError(f'This is not a valid expression: {expression}') return expression
[docs]class SelectedExpressionsIterator: """A multiple expression is an expression that contains Catalog. This iterator loops on pre-specified configurations """
[docs] def __init__(self, the_expression, configurations): """Ctor. :param the_expression: expression containing Catalogs :type the_expression: Expression :param configurations: set of configurations :type configurations: set(biogeme.configuration.Configuration)) """ self.the_expression = the_expression self.configurations = configurations self.set_iterator = iter(configurations) current_configuration = next(self.set_iterator) self.the_expression.configure_catalogs(current_configuration) self.first = True self.number = 0
def __iter__(self): return self def __next__(self): self.number += 1 if self.first: self.first = False return self.the_expression current_configuration = next(self.set_iterator) self.the_expression.configure_catalogs(current_configuration) return self.the_expression
[docs]class Expression: """This is the general arithmetic expression in biogeme. It serves as a base class for concrete expressions. """
[docs] def __init__(self): """Constructor""" self.children = [] #: List of children expressions self.id_manager = None #: in charge of the IDs self.keep_id_manager = None #: a copy of the ID manager self.fixedBetaValues = None """values of the Beta that are not estimated """ self.numberOfDraws = None """number of draws for Monte Carlo integration """ self._row = None """Row of the database where the values of the variables are found """ self.cpp = ee.pyEvaluateOneExpression() """ Interface to the C++ implementation """ self.missingData = 99999 """ Value interpreted as missing data """ self.central_controller = None """ Central controller for the multiple expressions """
def __iter__(self): the_set = self.set_of_configurations() return SelectedExpressionsIterator(self, the_set)
[docs] def check_panel_trajectory(self): """Set of variables defined outside of 'PanelLikelihoodTrajectory' :return: List of names of variables :rtype: set(str) """ check_children = set( chain.from_iterable( [e.check_panel_trajectory() for e in self.get_children()] ) ) return check_children
[docs] def check_draws(self): """Set of draws defined outside of 'MonteCarlo' :return: List of names of variables :rtype: set(str) """ check_children = set( chain.from_iterable([e.check_draws() for e in self.get_children()]) ) return check_children
[docs] def check_rv(self): """Set of random variables defined outside of 'Integrate' :return: List of names of variables :rtype: set(str) """ check_children = set( chain.from_iterable([e.check_rv() for e in self.get_children()]) ) return check_children
[docs] def getStatusIdManager(self): """Check the elementary expressions that are associated with an ID manager. :return: two sets of elementary expressions, those with and without an ID manager. :rtype: tuple(set(str), set(str)) """ with_id = set() without_id = set() for e in self.get_children(): yes, no = e.getStatusIdManager() with_id.update(yes) without_id.update(no) return with_id, without_id
[docs] def prepare(self, database, numberOfDraws): """Prepare the expression to be evaluated :param database: Biogeme database :type database: biogeme.database.Database :param numberOfDraws: number of draws for Monte-Carlo integration :type numberOfDraws: int """ # First, we reset the IDs, if any self.setIdManager(None) # Second, we calculate a new set of IDs. id_manager = IdManager([self], database, numberOfDraws) self.setIdManager(id_manager)
[docs] def setIdManager(self, id_manager): """The ID manager contains the IDs of the elementary expressions. It is externally created, as it may nee to coordinate the numbering of several expressions. It is stored only in the expressions of type Elementary. :param id_manager: ID manager to be propagated to the elementary expressions. If None, all the IDs are set to None. :type id_manager: class IdManager """ self.id_manager = id_manager for e in self.get_children(): e.setIdManager(id_manager)
def __repr__(self): """built-in function used to compute the 'official' string reputation of an object :return: description of the expression :rtype: string """ return self.__str__()
[docs] def __add__(self, other): """ Operator overloading. Generate an expression for addition. :param other: expression to be added :type other: biogeme.expressions.Expression :return: self + other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during addition to {self}: [{other}]' raise excep.BiogemeError(error_msg) return Plus(self, other)
[docs] def __radd__(self, other): """ Operator overloading. Generate an expression for addition. :param other: expression to be added :type other: biogeme.expressions.Expression :return: other + self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during addition to {self}: [{other}]' raise excep.BiogemeError(error_msg) return Plus(other, self)
[docs] def __sub__(self, other): """ Operator overloading. Generate an expression for substraction. :param other: expression to substract :type other: biogeme.expressions.Expression :return: self - other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during substraction to {self}: [{other}]' raise excep.BiogemeError(error_msg) return Minus(self, other)
[docs] def __rsub__(self, other): """ Operator overloading. Generate an expression for substraction. :param other: expression to be substracted :type other: biogeme.expressions.Expression :return: other - self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during substraction of {self}: [{other}]' raise excep.BiogemeError(error_msg) return Minus(other, self)
[docs] def __mul__(self, other): """ Operator overloading. Generate an expression for multiplication. :param other: expression to be multiplied :type other: biogeme.expressions.Expression :return: self * other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = ( f'Invalid expression during multiplication ' f'to {self}: [{other}]' ) raise excep.BiogemeError(error_msg) return Times(self, other)
[docs] def __rmul__(self, other): """ Operator overloading. Generate an expression for multiplication. :param other: expression to be multiplied :type other: biogeme.expressions.Expression :return: other * self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = ( f'Invalid expression during multiplication ' f'to {self}: [{other}]' ) raise excep.BiogemeError(error_msg) return Times(other, self)
[docs] def __div__(self, other): """ Operator overloading. Generate an expression for division. :param other: expression for division :type other: biogeme.expressions.Expression :return: self / other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during division of {self}: [{other}]' raise excep.BiogemeError(error_msg) return Divide(self, other)
[docs] def __rdiv__(self, other): """ Operator overloading. Generate an expression for division. :param other: expression for division :type other: biogeme.expressions.Expression :return: other / self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during division by {self}: [{other}]' raise excep.BiogemeError(error_msg) return Divide(other, self)
[docs] def __truediv__(self, other): """ Operator overloading. Generate an expression for division. :param other: expression for division :type other: biogeme.expressions.Expression :return: self / other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during division of {self}: [{other}]' raise excep.BiogemeError(error_msg) return Divide(self, other)
[docs] def __rtruediv__(self, other): """ Operator overloading. Generate an expression for division. :param other: expression for division :type other: biogeme.expressions.Expression :return: other / self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): error_msg = f'Invalid expression during division by {self}: [{other}]' raise excep.BiogemeError(error_msg) return Divide(other, self)
[docs] def __neg__(self): """ Operator overloading. Generate an expression for unary minus. :return: -self :rtype: biogeme.expressions.Expression """ return UnaryMinus(self)
[docs] def __pow__(self, other): """ Operator overloading. Generate an expression for power. :param other: expression for power :type other: biogeme.expressions.Expression :return: self ^ other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Power(self, other)
[docs] def __rpow__(self, other): """ Operator overloading. Generate an expression for power. :param other: expression for power :type other: biogeme.expressions.Expression :return: other ^ self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Power(other, self)
[docs] def __and__(self, other): """ Operator overloading. Generate an expression for logical and. :param other: expression for logical and :type other: biogeme.expressions.Expression :return: self and other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return And(self, other)
def __rand__(self, other): """ Operator overloading. Generate an expression for logical and. :param other: expression for logical and :type other: biogeme.expressions.Expression :return: other and self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return And(other, self)
[docs] def __or__(self, other): """ Operator overloading. Generate an expression for logical or. :param other: expression for logical or :type other: biogeme.expressions.Expression :return: self or other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Or(self, other)
def __ror__(self, other): """ Operator overloading. Generate an expression for logical or. :param other: expression for logical or :type other: biogeme.expressions.Expression :return: other or self :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Or(other, self)
[docs] def __eq__(self, other): """ Operator overloading. Generate an expression for comparison. :param other: expression for equality :type other: biogeme.expressions.Expression :return: self == other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Equal(self, other)
[docs] def __ne__(self, other): """ Operator overloading. Generate an expression for comparison. :param other: expression for difference :type other: biogeme.expressions.Expression :return: self != other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return NotEqual(self, other)
[docs] def __le__(self, other): """ Operator overloading. Generate an expression for comparison. :param other: expression for less or equal :type other: biogeme.expressions.Expression :return: self <= other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return LessOrEqual(self, other)
[docs] def __ge__(self, other): """ Operator overloading. Generate an expression for comparison. :param other: expression for greater or equal :type other: biogeme.expressions.Expression :return: self >= other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return GreaterOrEqual(self, other)
[docs] def __lt__(self, other): """ Operator overloading. Generate an expression for comparison. :param other: expression for less than :type other: biogeme.expressions.Expression :return: self < other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Less(self, other)
[docs] def __gt__(self, other): """ Operator overloading. Generate an expression for comparison. :param other: expression for greater than :type other: biogeme.expressions.Expression :return: self > other :rtype: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ if not (isNumeric(other) or isinstance(other, Expression)): raise excep.BiogemeError(f'This is not a valid expression: {other}') return Greater(self, other)
[docs] def createFunction( self, database=None, numberOfDraws=1000, gradient=True, hessian=True, bhhh=False, ): """Create a function based on the expression. The function takes as argument an array for the free parameters, and return the value of the function, the gradient, the hessian and the BHHH. The calculation of the derivatives is optional. :param database: database. If no database is provided, the expression must not contain any variable. :type database: biogeme.database.Database :param numberOfDraws: number of draws if needed by Monte-Carlo integration. :type numberOfDraws: int :param gradient: if True, the gradient is calculated. :type gradient: bool :param hessian: if True, the hessian is calculated. :type hessian: bool :param bhhh: if True, the BHHH matrix is calculated. :type bhhh: bool :return: the function. It will return, in that order, the value of the function, the gradient, the hessian and the BHHH matrix. Only requested quantities will be returned. For instance, if the gradient and the BHHH matrix are requested, and not the hessian, the tuple that is returned is f, g, bhhh. :rtype: fct(np.array) :raise BiogemeError: if gradient is False and hessian or BHHH is True. """ if (hessian or bhhh) and not gradient: raise excep.BiogemeError( 'If the hessian or BHHH is calculated, so is the gradient. ' 'The provided parameters are inconsistent.' ) with_id, without_id = self.getStatusIdManager() if len(without_id) > 0: if len(with_id) > 0: error_msg = ( f'IDs are defined for some expressions ' f'[{with_id}] but not for some [{without_id}]' ) raise excep.BiogemeError(error_msg) self.setIdManager(IdManager([self], database, numberOfDraws)) def my_function(x): if isinstance(x, (float, int, np.float64)): x = [float(x)] if len(x) != len(self.id_manager.free_betas_values): error_msg = ( f'Function is expecting an array of length ' f'{len(self.id_manager.free_betas_values)}, not {len(x)}' ) raise excep.BiogemeError(error_msg) self.id_manager.free_betas_values = x f, g, h, b = self.getValueAndDerivatives( database=database, numberOfDraws=numberOfDraws, gradient=gradient, hessian=hessian, bhhh=bhhh, aggregation=True, prepareIds=False, ) results = [f] if gradient: results.append(g) if hessian: results.append(h) if bhhh: results.append(b) return tuple(results) return f return my_function
[docs] def getValue_c( self, database=None, betas=None, numberOfDraws=1000, aggregation=False, prepareIds=False, ): """Evaluation of the expression, without the derivatives :param betas: values of the free parameters :type betas: list(float) :param database: database. If no database is provided, the expression must not contain any variable. :type database: biogeme.database.Database :param numberOfDraws: number of draws if needed by Monte-Carlo integration. :type numberOfDraws: int :param aggregation: if a database is provided, and this parameter is True, the expression is applied on each entry of the database, and all values are aggregated, so that the sum is returned. If False, the list of all values is returned. :type aggregation: bool :param prepareIds: if True, it means that the IDs of the expression must be constructed before the evaluation of the expression. :type prepareIds: bool :return: if a database is provided, a list where each entry is the result of applying the expression on one entry of the dsatabase. It returns a float. :rtype: np.array or float :raise BiogemeError: if no database is given, and the number of returned values is different from one. """ if self.requiresDraws() and database is None: error_msg = ( 'An expression involving MonteCarlo integration ' 'must be associated with a database.' ) raise excep.BiogemeError(error_msg) f, _, _, _ = self.getValueAndDerivatives( betas=betas, database=database, numberOfDraws=numberOfDraws, gradient=False, hessian=False, bhhh=False, aggregation=aggregation, prepareIds=prepareIds, ) if database is None: if len(f) != 1: error_msg = 'Incorrect number of return values' raise excep.BiogemeError(error_msg) return f[0] return f
[docs] def getValueAndDerivatives( self, betas=None, database=None, numberOfDraws=1000, gradient=True, hessian=True, bhhh=True, aggregation=True, prepareIds=False, ): """Evaluation of the expression In Biogeme the complexity of some expressions requires a specific implementation, in C++. This function invokes the C++ code to evaluate the value of the expression for a series of entries in a database. Note that this function will generate draws if needed. :param betas: values of the free parameters :type betas: list(float) :param database: database. If no database is provided, the expression must not contain any variable. :type database: biogeme.database.Database :param numberOfDraws: number of draws if needed by Monte-Carlo integration. :type numberOfDraws: int :param gradient: If True, the gradient is calculated. :type gradient: bool :param hessian: if True, the hessian is calculated. :type hessian: bool :param bhhh: if True, the BHHH matrix is calculated. :type bhhh: bool :param aggregation: if a database is provided, and this parameter is True, the expression is applied on each entry of the database, and all values are aggregated, so that the sum is returned. If False, the list of all values is returned. :type aggregation: bool :param prepareIds: if True, it means that the IDs of the expression must be constructed before the evaluation of the expression. :type prepareIds: bool :return: if a database is provided, a list where each entry is the result of applying the expression on one entry of the dsatabase. It returns a float, a vector, and a matrix, depedending if derivatives are requested. :rtype: np.array or float, numpy.array, numpy.array :raise BiogemeError: if no database is given and the expressions involves variables. :raise BiogemeError: if gradient is False and hessian or BHHH is True. :raise BiogemeError: if derivatives are asked, and the expression is not simple. :raise BiogemeError: if the expression involves MonteCarlo integration, and no database is provided. """ if prepareIds: self.keep_id_manager = self.id_manager self.prepare(database, numberOfDraws) elif self.id_manager is None: error_msg = 'Expression evaluated out of context. Set prepareIds to True.' raise excep.BiogemeError(error_msg) errors, warnings = self.audit(database) if warnings: logger.warning('\n'.join(warnings)) if errors: error_msg = '\n'.join(errors) logger.warning(error_msg) raise excep.BiogemeError(error_msg) if (hessian or bhhh) and not gradient: raise excep.BiogemeError( 'If the hessian or the BHHH matrix is calculated, ' 'so is the gradient. The provided parameters are inconsistent.' ) if database is None: variables = self.set_of_elementary_expression( TypeOfElementaryExpression.VARIABLE ) if variables: raise excep.BiogemeError( f'No database is provided and the expression ' f'contains variables: {variables}' ) self.numberOfDraws = numberOfDraws if database is not None: self.cpp.setData(database.data) if self.embedExpression('PanelLikelihoodTrajectory'): if database.isPanel(): database.buildPanelMap() self.cpp.setDataMap(database.individualMap) else: error_msg = ( 'The expression involves ' '"PanelLikelihoodTrajectory" ' 'that requires panel data' ) raise excep.BiogemeError(error_msg) if betas is not None: self.id_manager.free_betas_values = [ betas[x] if x in betas else self.id_manager.free_betas.expressions[x].initValue for x in self.id_manager.free_betas.names ] # List of values of the fixed beta parameters (those not estimated) self.fixedBetaValues = [ betas[x] if x in betas else self.id_manager.fixed_betas.expressions[x].initValue for x in self.id_manager.fixed_betas.names ] self.cpp.setExpression(self.getSignature()) self.cpp.setFreeBetas(self.id_manager.free_betas_values) self.cpp.setFixedBetas(self.id_manager.fixed_betas_values) self.cpp.setMissingData(self.missingData) if self.requiresDraws(): if database is None: error_msg = ( 'An expression involving MonteCarlo integration ' 'must be associated with a database.' ) raise excep.BiogemeError(error_msg) self.cpp.setDraws(database.theDraws) self.cpp.calculate( gradient=gradient, hessian=hessian, bhhh=bhhh, aggregation=aggregation, ) f, g, h, b = self.cpp.getResults() gres = g if gradient else None hres = h if hessian else None bhhhres = b if bhhh else None if aggregation: results = ( f[0], None if gres is None else g[0], None if hres is None else h[0], None if bhhhres is None else b[0], ) else: results = (f, gres, hres, bhhhres) # Now, if we had to set the IDS, we reset them as they cannot # be used in another context. if prepareIds: # We restore the previous Id manager self.setIdManager(self.keep_id_manager) return results
[docs] def requiresDraws(self): """Checks if the expression requires draws :return: True if it requires draws. :rtype: bool """ return self.embedExpression('MonteCarlo')
[docs] def get_beta_values(self): """Returns a dict with the initial values of beta. Typically useful for simulation. :return: dict with the initial values of the beta :rtype: dict(str: float) """ betas = self.dict_of_elementary_expression(TypeOfElementaryExpression.FREE_BETA) return {b.name: b.initValue for b in betas.values()}
[docs] def set_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a specific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression :return: returns a set with the names of the elementary expressions :rtype: set(string.Expression) """ return set(self.dict_of_elementary_expression(the_type).keys())
[docs] def dict_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a specific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression :return: returns a dict with the variables appearing in the expression the keys being their names. :rtype: dict(string:biogeme.expressions.Expression) """ return dict( chain( *( e.dict_of_elementary_expression(the_type).items() for e in self.children ) ) )
[docs] def getElementaryExpression(self, name): """Return: an elementary expression from its name if it appears in the expression. :param name: name of the elementary expression. :type name: string :return: the expression if it exists. None otherwise. :rtype: biogeme.expressions.Expression """ for e in self.get_children(): if e.getElementaryExpression(name) is not None: return e.getElementaryExpression(name) return None
[docs] def setRow(self, row): """Obsolete function. This function identifies the row of the database from which the values of the variables must be obtained. :param row: row from the database :type row: pandas.core.series.Serie :raise BiogemeError: if the function is called, because it is obsolete. """ raise excep.BiogemeError("The function setRow is now obsolete.")
[docs] def rename_elementary(self, names, prefix=None, suffix=None): """Rename elementary expressions by adding a prefix and/or a suffix :param names: names of expressions to rename :type names: list(str) :param prefix: if not None, the expression is renamed, with a prefix defined by this argument. :type prefix: str :param suffix: if not None, the expression is renamed, with a suffix defined by this argument. :type suffix: str """ for e in self.get_children(): e.rename_elementary(names, prefix=prefix, suffix=suffix)
[docs] def fix_betas(self, beta_values, prefix=None, suffix=None): """Fix all the values of the beta parameters appearing in the dictionary :param beta_values: dictionary containing the betas to be fixed (as key) and their value. :type beta_values: dict(str: float) :param prefix: if not None, the parameter is renamed, with a prefix defined by this argument. :type prefix: str :param suffix: if not None, the parameter is renamed, with a suffix defined by this argument. :type suffix: str """ for e in self.get_children(): e.fix_betas(beta_values, prefix=prefix, suffix=suffix)
[docs] def getClassName(self): """ Obtain the name of the top class of the expression structure :return: the name of the class :rtype: string """ n = type(self).__name__ return n
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the signatures of all the children expressions, 2. the name of the expression between < > 3. the id of the expression between { } 4. the number of children between ( ) 5. the ids of each children, preceeded by a comma. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) } { \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ listOfSignatures = [] for e in self.get_children(): listOfSignatures += e.getSignature() mysignature = f'<{self.getClassName()}>' mysignature += f'{{{self.get_id()}}}' mysignature += f'({len(self.get_children())})' for e in self.get_children(): mysignature += f',{e.get_id()}' listOfSignatures += [mysignature.encode()] return listOfSignatures
[docs] def embedExpression(self, t): """Check if the expression contains an expression of type t. Typically, this would be used to check that a MonteCarlo expression contains a bioDraws expression. :return: True if the expression contains an expression of type t. :rtype: bool """ if self.getClassName() == t: return True for e in self.get_children(): if e.embedExpression(t): return True return False
[docs] def countPanelTrajectoryExpressions(self): """Count the number of times the PanelLikelihoodTrajectory is used in the formula. It should trigger an error if it is used more than once. :return: number of times the PanelLikelihoodTrajectory is used in the formula :rtype: int """ nbr = 0 for e in self.get_children(): nbr += e.countPanelTrajectoryExpressions() return nbr
[docs] def audit(self, database=None): """Performs various checks on the expressions. :param database: database object :type database: biogeme.database.Database :return: tuple list_of_errors, list_of_warnings :rtype: list(string), list(string) """ list_of_errors = [] list_of_warnings = [] for e in self.get_children(): if not isinstance(e, Expression): the_error = f'Invalid expression: {e}' list_of_errors.append(the_error) err, war = e.audit(database) list_of_errors += err list_of_warnings += war return list_of_errors, list_of_warnings
[docs] def change_init_values(self, betas): """Modifies the initial values of the Beta parameters. The fact that the parameters are fixed or free is irrelevant here. :param betas: dictionary where the keys are the names of the parameters, and the values are the new value for the parameters. :type betas: dict(string:float) """ for e in self.get_children(): e.change_init_values(betas)
[docs] def dict_of_catalogs(self, ignore_synchronized=False): """Returns a dict with all catalogs in the expression. :return: dict with all the catalogs """ result = {} for e in self.children: a_dict = e.dict_of_catalogs(ignore_synchronized) for key, the_catalog in a_dict.items(): result[key] = the_catalog return result
[docs] def contains_catalog(self, name): """Check if the expression contains a specific catalog :param name: name of the catalog to search. :type name: str :return: True if the given catalog is contained in the expression. False otherwise. :rtype: bool """ all_catalogs = self.dict_of_catalogs() return name in all_catalogs
[docs] def set_central_controller(self, the_central_controller=None): if the_central_controller is None: self.central_controller = CentralController( expression=self, ) else: self.central_controller = the_central_controller for e in self.children: e.set_central_controller(self.central_controller) return self.central_controller
[docs] def get_all_controllers(self): """Provides all controllers controlling the specifications of a multiple expression :return: a set of controllers :rtype: set(biogeme.controller.Controller) """ if not self.children: return set() all_controllers = set() for e in self.children: all_controllers |= e.get_all_controllers() return all_controllers
[docs] def number_of_multiple_expressions(self): """Reports the number of multiple expressions available through the iterator :return: number of multiple expressions :rtype: int """ if self.central_controller is None: self.set_central_controller() return self.central_controller.number_of_configurations()
[docs] def set_of_configurations(self): """Provides the set of all possible configurations""" if self.central_controller is None: self.set_central_controller() return self.central_controller.all_configurations
[docs] def reset_expression_selection(self): """In each group of expressions, select the first one""" for e in self.children: e.reset_expression_selection()
[docs] def configure_catalogs(self, configuration): """Select the items in each catalog corresponding to the requested configuration :param configuration: catalog configuration :type configuration: biogeme.configuration.Configuration """ if self.central_controller is None: self.set_central_controller() self.central_controller.set_configuration(configuration)
[docs] def current_configuration(self): """Obtain the current configuration of an expression :return: configuration :rtype: biogeme.configuration.Configuration """ if self.central_controller is None: self.set_central_controller() return self.central_controller.get_configuration()
[docs] def select_expression(self, controller_name, index): """Select a specific expression in a group :param controller_name: name of the controller :type controller_name: str :param index: index of the expression in the group :type index: int :raises BiogemeError: if index is out of range """ if self.central_controller is None: self.set_central_controller() self.central_controller.set_controller(controller_name, index)
[docs] def set_of_multiple_expressions(self): """Set of the multiple expressions found in the current expression :return: a set of descriptions of the multiple expressions :rtype: set(MultipleExpressionDescription) """ all_sets = [e.set_of_multiple_expressions() for e in self.get_children()] return set(chain.from_iterable(all_sets))
[docs] def get_id(self): """Retrieve the id of the expression used in the signature :return: id of the object :rtype: int """ return id(self)
[docs] def get_children(self): """Retrieve the list of children :return: list of children :rtype: list(Expression) """ return self.children
[docs]class BinaryOperator(Expression): """ Base class for arithmetic expressions that are binary operators. This expression is the result of the combination of two expressions, typically addition, substraction, multiplication or division. """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ Expression.__init__(self) if isNumeric(left): self.left = Numeric(left) #: left child else: if not isinstance(left, Expression): raise excep.BiogemeError(f'This is not a valid expression: {left}') self.left = left if isNumeric(right): self.right = Numeric(right) #: right child else: if not isinstance(right, Expression): raise excep.BiogemeError(f'This is not a valid expression: {right}') self.right = right self.children.append(self.left) self.children.append(self.right)
[docs]class Plus(BinaryOperator): """ Addition expression """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} + {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.left.getValue() + self.right.getValue()
[docs]class Minus(BinaryOperator): """ Substraction expression """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} - {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.left.getValue() - self.right.getValue()
[docs]class Times(BinaryOperator): """ Multiplication expression """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} * {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.left.getValue() * self.right.getValue()
[docs]class Divide(BinaryOperator): """ Division expression """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} / {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.left.getValue() / self.right.getValue()
[docs]class Power(BinaryOperator): """ Power expression """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} ** {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.left.getValue() ** self.right.getValue()
[docs]class bioMin(BinaryOperator): """ Minimum of two expressions """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'bioMin({self.left}, {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ if self.left.getValue() <= self.right.getValue(): return self.left.getValue() return self.right.getValue()
[docs]class bioMax(BinaryOperator): """ Maximum of two expressions """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'bioMax({self.left}, {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ if self.left.getValue() >= self.right.getValue(): return self.left.getValue() return self.right.getValue()
[docs]class And(BinaryOperator): """ Logical and """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} and {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ if self.left.getValue() == 0.0: return 0.0 if self.right.getValue() == 0.0: return 0.0 return 1.0
[docs]class Or(BinaryOperator): """ Logical or """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} or {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ if self.left.getValue() != 0.0: return 1.0 if self.right.getValue() != 0.0: return 1.0 return 0.0
[docs]class ComparisonOperator(BinaryOperator): """Base class for comparison expressions."""
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ BinaryOperator.__init__(self, left, right)
[docs] def audit(self, database=None): """Performs various checks on the expression.""" list_of_errors = [] list_of_warnings = [] if isinstance(self.left, ComparisonOperator) or isinstance( self.right, ComparisonOperator ): the_warning = ( f'The following expression may potentially be ambiguous: [{self}] ' f'if it contains the chaining of two comparisons expressions. ' f'Keep in mind that, for Biogeme (like for Pandas), the ' f'expression (a <= x <= b) is not equivalent to (a <= x) ' f'and (x <= b).' ) list_of_warnings.append(the_warning) return list_of_errors, list_of_warnings
[docs]class Equal(ComparisonOperator): """ Logical equal """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ ComparisonOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} == {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ r = 1 if self.left.getValue() == self.right.getValue() else 0 return r
[docs]class NotEqual(ComparisonOperator): """ Logical not equal """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ ComparisonOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} != {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ r = 1 if self.left.getValue() != self.right.getValue() else 0 return r
[docs]class LessOrEqual(ComparisonOperator): """ Logical less or equal """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ ComparisonOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} <= {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ r = 1 if self.left.getValue() <= self.right.getValue() else 0 return r
[docs]class GreaterOrEqual(ComparisonOperator): """ Logical greater or equal """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ ComparisonOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} >= {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ r = 1 if self.left.getValue() >= self.right.getValue() else 0 return r
[docs]class Less(ComparisonOperator): """ Logical less """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ ComparisonOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} < {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ r = 1 if self.left.getValue() < self.right.getValue() else 0 return r
[docs]class Greater(ComparisonOperator): """ Logical greater """
[docs] def __init__(self, left, right): """Constructor :param left: first arithmetic expression :type left: biogeme.expressions.Expression :param right: second arithmetic expression :type right: biogeme.expressions.Expression """ ComparisonOperator.__init__(self, left, right)
def __str__(self): return f'({self.left} > {self.right})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ r = 1 if self.left.getValue() > self.right.getValue() else 0 return r
[docs]class UnaryOperator(Expression): """ Base class for arithmetic expressions that are unary operators. Such an expression is the result of the modification of another expressions, typically changing its sign. """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ Expression.__init__(self) if isNumeric(child): self.child = Numeric(child) #: child else: if not isinstance(child, Expression): raise excep.BiogemeError(f'This is not a valid expression: {child}') self.child = child self.children.append(self.child)
[docs]class UnaryMinus(UnaryOperator): """ Unary minus expression """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'(-{self.child})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return -self.child.getValue()
[docs]class MonteCarlo(UnaryOperator): """ Monte Carlo integration """
[docs] def __init__(self, child): """Constructor :param child: arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'MonteCarlo({self.child})'
[docs] def check_draws(self): """List of draws defined outside of 'MonteCarlo' :return: List of names of variables :rtype: list(str) """ return set()
[docs] def audit(self, database=None): """Performs various checks on the expressions. :param database: database object :type database: biogeme.database.Database :return: tuple list_of_errors, list_of_warnings :rtype: list(string), list(string) """ list_of_errors, list_of_warnings = self.child.audit(database) if database is None: if self.child.embedExpression('PanelLikelihoodTrajectory'): theWarning = ( 'The formula contains a PanelLikelihoodTrajectory ' 'expression, and no database is given' ) list_of_warnings.append(theWarning) else: if database.isPanel() and not self.child.embedExpression( 'PanelLikelihoodTrajectory' ): the_error = ( f'As the database is panel, the argument ' f'of MonteCarlo must contain a' f' PanelLikelihoodTrajectory: {self}' ) list_of_errors.append(the_error) if not self.child.embedExpression('bioDraws'): the_error = ( f'The argument of MonteCarlo must contain a' f' bioDraws: {self}' ) list_of_errors.append(the_error) if self.child.embedExpression('MonteCarlo'): the_error = ( f'It is not possible to include a MonteCarlo ' f'statement in another one: {self}' ) list_of_errors.append(the_error) return list_of_errors, list_of_warnings
[docs]class bioNormalCdf(UnaryOperator): """ Cumulative Distribution Function of a normal random variable """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'bioNormalCdf({self.child})'
[docs]class PanelLikelihoodTrajectory(UnaryOperator): """ Likelihood of a sequences of observations for the same individual """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'PanelLikelihoodTrajectory({self.child})'
[docs] def check_panel_trajectory(self): """List of variables defined outside of 'PanelLikelihoodTrajectory' :return: List of names of variables :rtype: list(str) """ return set()
[docs] def countPanelTrajectoryExpressions(self): """Count the number of times the PanelLikelihoodTrajectory is used in the formula. """ return 1 + self.child.countPanelTrajectoryExpressions()
[docs] def audit(self, database=None): """Performs various checks on the expressions. :param database: database object :type database: biogeme.database.Database :return: tuple list_of_errors, list_of_warnings :rtype: list(string), list(string) """ list_of_errors, list_of_warnings = self.child.audit(database) if not database.isPanel(): the_error = ( f'Expression PanelLikelihoodTrajectory can ' f'only be used with panel data. Use the statement ' f'database.panel("IndividualId") to declare the ' f'panel structure of the data: {self}' ) list_of_errors.append(the_error) return list_of_errors, list_of_warnings
[docs]class exp(UnaryOperator): """ exponential expression """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'exp({self.child})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return np.exp(self.child.getValue())
[docs]class log(UnaryOperator): """ logarithm expression """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'log({self.child})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return np.log(self.child.getValue())
[docs]class logzero(UnaryOperator): """ logarithm expression. Returns zero if the argument is zero. """
[docs] def __init__(self, child): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child)
def __str__(self): return f'logzero({self.child})'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ v = self.child.getValue() return 0 if v == 0 else np.log(v)
[docs]class Derive(UnaryOperator): """ Derivative with respect to an elementary expression """
[docs] def __init__(self, child, name): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression """ UnaryOperator.__init__(self, child) # Name of the elementary expression by which the derivative is taken self.elementaryName = name
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the signatures of the child expression, 2. the name of the expression between < > 3. the id of the expression between { } 4. the id of the child, preceeded by a comma. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ elementaryIndex = self.id_manager.elementary_expressions.indices[ self.elementaryName ] listOfSignatures = [] listOfSignatures += self.child.getSignature() mysignature = f'<{self.getClassName()}>' mysignature += f'{{{self.get_id()}}}' mysignature += f',{self.child.get_id()}' mysignature += f',{elementaryIndex}' listOfSignatures += [mysignature.encode()] return listOfSignatures
def __str__(self): return 'Derive({self.child}, "{self.elementName}")'
[docs]class Integrate(UnaryOperator): """ Numerical integration """
[docs] def __init__(self, child, name): """Constructor :param child: first arithmetic expression :type child: biogeme.expressions.Expression :param name: name of the random variable for the integration. :type name: string """ UnaryOperator.__init__(self, child) self.randomVariableName = name
[docs] def check_rv(self): """List of random variables defined outside of 'Integrate' :return: List of names of variables :rtype: list(str) """ return set()
[docs] def audit(self, database=None): """Performs various checks on the expressions. :param database: database object :type database: biogeme.database.Database :return: tuple list_of_errors, list_of_warnings :rtype: list(string), list(string) """ list_of_errors, list_of_warnings = self.child.audit(database) if not self.child.embedExpression('RandomVariable'): the_error = ( f'The argument of Integrate must contain a ' f'RandomVariable: {self}' ) list_of_errors.append(the_error) return list_of_errors, list_of_warnings
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the signatures of the child expression, 2. the name of the expression between < > 3. the id of the expression between { }, preceeded by a comma 4. the id of the children, preceeded by a comma 5. the index of the randon variable, preceeded by a comma Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) } { \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ randomVariableIndex = self.id_manager.random_variables.indices[ self.randomVariableName ] listOfSignatures = [] listOfSignatures += self.child.getSignature() mysignature = f'<{self.getClassName()}>' mysignature += f'{{{self.get_id()}}}' mysignature += f',{self.child.get_id()}' mysignature += f',{randomVariableIndex}' listOfSignatures += [mysignature.encode()] return listOfSignatures
def __str__(self): return f'Integrate({self.child}, "{self.randomVariableName}")'
[docs]class Elementary(Expression): """Elementary expression. It is typically defined by a name appearing in an expression. It can be a variable (from the database), or a parameter (fixed or to be estimated using maximum likelihood), a random variable for numrerical integration, or Monte-Carlo integration. """
[docs] def __init__(self, name): """Constructor :param name: name of the elementary experession. :type name: string """ Expression.__init__(self) self.name = name #: name of the elementary expressiom self.elementaryIndex = None """The id should be unique for all elementary expressions appearing in a given set of formulas. """
def __str__(self): """string method :return: name of the expression :rtype: str """ return f'{self.name}'
[docs] def getStatusIdManager(self): """Check the elementary expressions that are associated with an ID manager. :return: two lists of elementary expressions, those with and without an ID manager. :rtype: tuple(list(str), list(str)) """ if self.id_manager is None: return [], [self.name] return [self.name], []
[docs] def getElementaryExpression(self, name): """ :return: an elementary expression from its name if it appears in the expression. None otherwise. :rtype: biogeme.Expression """ if self.name == name: return self return None
[docs] def rename_elementary(self, names, prefix=None, suffix=None): """Rename elementary expressions by adding a prefix and/or a suffix :param names: names of expressions to rename :type names: list(str) :param prefix: if not None, the expression is renamed, with a prefix defined by this argument. :type prefix: str :param suffix: if not None, the expression is renamed, with a suffix defined by this argument. :type suffix: str """ if self.name in names: if prefix is not None: self.name = f'{prefix}{self.name}' if suffix is not None: self.name = f'{self.name}{suffix}'
[docs] def number_of_multiple_expressions(self): """Count the number of "parallel" expressions :return: the number of expressions :rtype: int """ return 1
[docs]class bioDraws(Elementary): """ Draws for Monte-Carlo integration """
[docs] def __init__(self, name, drawType): """Constructor :param name: name of the random variable with a series of draws. :type name: string :param drawType: type of draws. :type drawType: string """ Elementary.__init__(self, name) self.drawType = drawType self.drawId = None
def __str__(self): return f'bioDraws("{self.name}", "{self.drawType}")'
[docs] def check_draws(self): """List of draws defined outside of 'MonteCarlo' :return: List of names of variables :rtype: list(str) """ return {self.name}
[docs] def setIdManager(self, id_manager=None): """The ID manager contains the IDs of the elementary expressions. It is externally created, as it may nee to coordinate the numbering of several expressions. It is stored only in the expressions of type Elementary. :param id_manager: ID manager to be propagated to the elementary expressions. If None, all the IDs are set to None. :type id_manager: class IdManager """ self.id_manager = id_manager if id_manager is None: self.elementaryIndex = None self.drawId = None return self.elementaryIndex = self.id_manager.elementary_expressions.indices[self.name] self.drawId = self.id_manager.draws.indices[self.name]
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the name of the expression between < > 2. the id of the expression between { }, preceeded by a comma 3. the name of the draws 4. the unique ID (preceeded by a comma), 5. the draw ID (preceeded by a comma). Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) :raise biogeme.exceptions.BiogemeError: if no id has been defined for elementary expression :raise biogeme.exceptions.BiogemeError: if no id has been defined for draw """ if self.elementaryIndex is None: error_msg = ( f'No id has been defined for elementary ' f'expression {self.name}.' ) raise excep.BiogemeError(error_msg) if self.drawId is None: error_msg = f'No id has been defined for draw {self.name}.' raise excep.BiogemeError(error_msg) signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f'"{self.name}",{self.elementaryIndex},{self.drawId}' return [signature.encode()]
[docs] def dict_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a dpecific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression """ if the_type == TypeOfElementaryExpression.DRAWS: return {self.name: self.drawType} return {}
[docs]class Numeric(Expression): """ Numerical expression for a simple number """
[docs] def __init__(self, value): """Constructor :param value: numerical value :type value: float """ Expression.__init__(self) self.value = float(value) #: numeric value
def __str__(self): return '`' + str(self.value) + '`'
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.value
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the name of the expression between < > 2. the id of the expression between { } 3. the value, preceeded by a comma. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f',{self.value}' return [signature.encode()]
[docs]class Variable(Elementary): """Explanatory variable This represents the explanatory variables of the choice model. Typically, they come from the data set. """
[docs] def __init__(self, name): """Constructor :param name: name of the variable. :type name: string """ Elementary.__init__(self, name) # Index of the variable self.variableId = None
[docs] def check_panel_trajectory(self): """List of variables defined outside of 'PanelLikelihoodTrajectory' :return: List of names of variables :rtype: list(str) """ return {self.name}
[docs] def getValue(self): """The evaluation of a Variable requires a database. Therefore, this function triggers an exception. :raise BiogemeError: each time the function is calles """ error_msg = ( f'Evaluating Variable {self.name} requires a database. Use the ' f'function getValue_c instead.' ) raise excep.BiogemeError(error_msg)
[docs] def setIdManager(self, id_manager=None): """The ID manager contains the IDs of the elementary expressions. It is externally created, as it may need to coordinate the numbering of several expressions. It is stored only in the expressions of type Elementary. :param id_manager: ID manager to be propagated to the elementary expressions. If None, all the IDs are set to None. :type id_manager: class IdManager """ self.id_manager = id_manager if id_manager is None: self.elementaryIndex = None self.variableId = None return self.elementaryIndex = self.id_manager.elementary_expressions.indices[self.name] self.variableId = self.id_manager.variables.indices[self.name]
[docs] def dict_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a specific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression """ if the_type == TypeOfElementaryExpression.VARIABLE: return {self.name: self} return {}
[docs] def audit(self, database=None): """Performs various checks on the expressions. :param database: database object :type database: biogeme.database.Database :return: tuple list_of_errors, list_of_warnings :rtype: list(string), list(string) :raise BiogemeError: if no database is provided. :raise BiogemeError: if the name of the variable does not appear in the database. """ list_of_errors = [] list_of_warnings = [] if database is None: raise excep.BiogemeError( 'The database must be provided to audit the variable.' ) if self.name not in database.data.columns: the_error = f'Variable {self.name} not found in the database.' list_of_errors.append(the_error) return list_of_errors, list_of_warnings
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the name of the expression between < > 2. the id of the expression between { } 3. the name of the variable, 4. the unique ID, preceeded by a comma. 5. the variabvle ID, preceeded by a comma. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) :raise biogeme.exceptions.BiogemeError: if no id has been defined for elementary expression :raise biogeme.exceptions.BiogemeError: if no id has been defined for variable """ if self.elementaryIndex is None: error_msg = ( f'No id has been defined for elementary expression ' f'{self.name}.' ) raise excep.BiogemeError(error_msg) if self.variableId is None: error_msg = f'No id has been defined for variable {self.name}.' raise excep.BiogemeError(error_msg) signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f'"{self.name}",{self.elementaryIndex},{self.variableId}' return [signature.encode()]
[docs]class DefineVariable(Variable): """Expression that defines a new variable and add a column in the database. This expression allows the use to define a new variable that will be added to the database. It avoids that it is recalculated each time it is needed. """
[docs] def __init__(self, name, expression, database): """Constructor :param name: name of the variable. :type name: string :param expression: formula that defines the variable :type expression: biogeme.expressions.Expression :param database: object identifying the database. :type database: biogeme.database.Database :raise BiogemeError: if the expression is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ raise excep.BiogemeError( 'This expression is obsolete. Use the same function in the ' 'database object. Replace "new_var = DefineVariable(\'NEW_VAR\',' ' expression, database)" by "new_var = database.DefineVariable' '(\'NEW_VAR\', expression)"' )
[docs]class RandomVariable(Elementary): """ Random variable for numerical integration """
[docs] def __init__(self, name): """Constructor :param name: name of the random variable involved in the integration. :type name: string. """ Elementary.__init__(self, name) # Index of the random variable self.rvId = None
[docs] def check_rv(self): """List of random variables defined outside of 'Integrate' :return: List of names of variables :rtype: list(str) """ return {self.name}
[docs] def setIdManager(self, id_manager=None): """The ID manager contains the IDs of the elementary expressions. It is externally created, as it may nee to coordinate the numbering of several expressions. It is stored only in the expressions of type Elementary. :param id_manager: ID manager to be propagated to the elementary expressions. If None, all the IDs are set to None. :type id_manager: class IdManager """ self.id_manager = id_manager if id_manager is None: self.elementaryIndex = None self.rvId = None return self.elementaryIndex = self.id_manager.elementary_expressions.indices[self.name] self.rvId = self.id_manager.random_variables.indices[self.name]
[docs] def dict_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a specific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression """ if the_type == TypeOfElementaryExpression.RANDOM_VARIABLE: return {self.name: self} return {}
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the name of the expression between < > 2. the id of the expression between { } 3. the name of the random variable, 4. the unique ID, preceeded by a comma, 5. the ID of the random variable. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) :raise biogeme.exceptions.BiogemeError: if no id has been defined for elementary expression :raise biogeme.exceptions.BiogemeError: if no id has been defined for random variable """ if self.elementaryIndex is None: error_msg = ( f'No id has been defined for elementary ' f'expression {self.name}.' ) raise excep.BiogemeError(error_msg) if self.rvId is None: error_msg = f'No id has been defined for random variable {self.name}.' raise excep.BiogemeError(error_msg) signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f'"{self.name}",{self.elementaryIndex},{self.rvId}' return [signature.encode()]
[docs]class Beta(Elementary): """ Unknown parameters to be estimated from data. """
[docs] def __init__(self, name, value, lowerbound, upperbound, status): """Constructor :param name: name of the parameter. :type name: string :param value: default value. :type value: float :param lowerbound: if different from None, imposes a lower bound on the value of the parameter during the optimization. :type lowerbound: float :param upperbound: if different from None, imposes an upper bound on the value of the parameter during the optimization. :type upperbound: float :param status: if different from 0, the parameter is fixed to its default value, and not modified by the optimization algorithm. :type status: int :raise BiogemeError: if the first parameter is not a str. :raise BiogemeError: if the second parameter is not a int or a float. """ if not isinstance(value, (int, float)): error_msg = ( f'The second parameter for {name} must be ' f'a float and not a {type(value)}: {value}' ) raise excep.BiogemeError(error_msg) if not isinstance(name, str): error_msg = ( f'The first parameter must be a string and ' f'not a {type(name)}: {name}' ) raise excep.BiogemeError(error_msg) Elementary.__init__(self, name) self.initValue = value self.lb = lowerbound self.ub = upperbound self.status = status self.betaId = None
[docs] def setIdManager(self, id_manager=None): """The ID manager contains the IDs of the elementary expressions. It is externally created, as it may need to coordinate the numbering of several expressions. It is stored only in the expressions of type Elementary. :param id_manager: ID manager to be propagated to the elementary expressions. If None, all the IDs are set to None. :type id_manager: class IdManager """ self.id_manager = id_manager if id_manager is None: self.elementaryIndex = None self.betaId = None return self.elementaryIndex = self.id_manager.elementary_expressions.indices[self.name] if self.status != 0: self.betaId = self.id_manager.fixed_betas.indices[self.name] else: self.betaId = self.id_manager.free_betas.indices[self.name]
def __str__(self): if self.status == 0: return f'{self.name}(init={self.initValue})' return f'{self.name}(fixed={self.initValue})'
[docs] def fix_betas(self, beta_values, prefix=None, suffix=None): """Fix all the values of the beta parameters appearing in the dictionary :param beta_values: dictionary containing the betas to be fixed (as key) and their value. :type beta_values: dict(str: float) :param prefix: if not None, the parameter is renamed, with a prefix defined by this argument. :type prefix: str :param suffix: if not None, the parameter is renamed, with a suffix defined by this argument. :type suffix: str """ if self.name in beta_values: self.initValue = beta_values[self.name] self.status = 1 if prefix is not None: self.name = f'{prefix}{self.name}' if suffix is not None: self.name = f'{self.name}{suffix}'
[docs] def dict_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a specific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression :return: returns a dict with the variables appearing in the expression the keys being their names. :rtype: dict(string:biogeme.expressions.Expression) """ if the_type == TypeOfElementaryExpression.BETA: return {self.name: self} if the_type == TypeOfElementaryExpression.FREE_BETA and self.status == 0: return {self.name: self} if the_type == TypeOfElementaryExpression.FIXED_BETA and self.status != 0: return {self.name: self} return {}
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ return self.initValue
[docs] def change_init_values(self, betas): """Modifies the initial values of the Beta parameters. The fact that the parameters are fixed or free is irrelevant here. :param betas: dictionary where the keys are the names of the parameters, and the values are the new value for the parameters. :type betas: dict(string:float) """ if self.name in betas: self.initValue = betas[self.name]
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the name of the expression between < > 2. the id of the expression between { } 3. the name of the parameter, 4. the status between [ ] 5. the unique ID, preceeded by a comma 6. the beta ID, preceeded by a comma Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) :raise biogeme.exceptions.BiogemeError: if no id has been defined for elementary expression :raise biogeme.exceptions.BiogemeError: if no id has been defined for parameter """ if self.elementaryIndex is None: error_msg = ( f'No id has been defined for elementary ' f'expression {self.name}.' ) raise excep.BiogemeError(error_msg) if self.betaId is None: raise excep.BiogemeError( f'No id has been defined for parameter {self.name}.' ) signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += ( f'"{self.name}"[{self.status}],' f'{self.elementaryIndex},{self.betaId}' ) return [signature.encode()]
[docs]class LogLogit(Expression): """Expression capturing the logit formula. It contains one formula for the target alternative, a dict of formula for the availabilities and a dict of formulas for the utilities """
[docs] def __init__(self, util, av, choice): """Constructor :param util: dictionary where the keys are the identifiers of the alternatives, and the elements are objects defining the utility functions. :type util: dict(int:biogeme.expressions.Expression) :param av: dictionary where the keys are the identifiers of the alternatives, and the elements are object of type biogeme.expressions.Expression defining the availability conditions. If av is None, all the alternatives are assumed to be always available :type av: dict(int:biogeme.expressions.Expression) :param choice: formula to obtain the alternative for which the logit probability must be calculated. :type choice: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ Expression.__init__(self) self.util = {} #: dict of utility functions for i, e in util.items(): if isNumeric(e): self.util[i] = Numeric(e) else: if not isinstance(e, Expression): raise excep.BiogemeError(f'This is not a valid expression: {e}') self.util[i] = e self.av = {} #: dict of availability formulas if av is None: self.av = {k: Numeric(1) for k, v in util.items()} else: for i, e in av.items(): if isNumeric(e): self.av[i] = Numeric(e) else: if not isinstance(e, Expression): raise excep.BiogemeError(f'This is not a valid expression: {e}') self.av[i] = e if isNumeric(choice): self.choice = Numeric(choice) """expression for the chosen alternative""" else: if not isinstance(choice, Expression): raise excep.BiogemeError(f'This is not a valid expression: {choice}') self.choice = choice self.children.append(self.choice) for i, e in self.util.items(): self.children.append(e) for i, e in self.av.items(): self.children.append(e)
[docs] def audit(self, database=None): """Performs various checks on the expressions. :param database: database object :type database: biogeme.database.Database :return: tuple list_of_errors, list_of_warnings :rtype: list(string), list(string) """ list_of_errors = [] list_of_warnings = [] for e in self.children: err, war = e.audit(database) list_of_errors += err list_of_warnings += war if self.util.keys() != self.av.keys(): the_error = 'Incompatible list of alternatives in logit expression. ' consistent = False myset = self.util.keys() - self.av.keys() if myset: mysetContent = ', '.join(f'{str(k)} ' for k in myset) the_error += ( 'Id(s) used for utilities and not for ' 'availabilities: ' ) + mysetContent myset = self.av.keys() - self.util.keys() if myset: mysetContent = ', '.join(f'{str(k)} ' for k in myset) the_error += ( ' Id(s) used for availabilities and not ' 'for utilities: ' ) + mysetContent list_of_errors.append(the_error) else: consistent = True listOfAlternatives = list(self.util) if database is None: choices = np.array([self.choice.getValue_c()]) else: choices = database.valuesFromDatabase(self.choice) correctChoices = np.isin(choices, listOfAlternatives) indexOfIncorrectChoices = np.argwhere(~correctChoices) if indexOfIncorrectChoices.any(): incorrectChoices = choices[indexOfIncorrectChoices] content = '-'.join( '{}[{}]'.format(*t) for t in zip(indexOfIncorrectChoices, incorrectChoices) ) truncate = 100 if len(content) > truncate: content = f'{content[:truncate]}...' the_error = ( f'The choice variable [{self.choice}] does not ' f'correspond to a valid alternative for the ' f'following observations (rownumber[choice]): ' ) + content list_of_errors.append(the_error) if consistent: if database is None: value_choice = self.choice.getValue_c() if value_choice not in self.av.keys(): the_error = ( f'The chosen alternative [{value_choice}] ' f'is not available' ) list_of_warnings.append(the_error) else: choiceAvailability = database.checkAvailabilityOfChosenAlt( self.av, self.choice ) indexOfUnavailableChoices = np.where(~choiceAvailability)[0] if indexOfUnavailableChoices.size > 0: incorrectChoices = choices[indexOfUnavailableChoices] content = '-'.join( '{}[{}]'.format(*t) for t in zip(indexOfUnavailableChoices, incorrectChoices) ) truncate = 100 if len(content) > truncate: content = f'{content[:truncate]}...' the_error = ( f'The chosen alternative [{self.choice}] ' f'is not available for the following ' f'observations (rownumber[choice]): ' ) + content list_of_warnings.append(the_error) return list_of_errors, list_of_warnings
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float :raise BiogemeError: if the chosen alternative does not correspond to any of the utility functions :raise BiogemeError: if the chosen alternative does not correspond to any of entry in the availability condition """ choice = int(self.choice.getValue()) if choice not in self.util: error_msg = ( f'Alternative {choice} does not appear in the list ' f'of utility functions: {self.util.keys()}' ) raise excep.BiogemeError(error_msg) if choice not in self.av: error_msg = ( f'Alternative {choice} does not appear in the list ' f'of availabilities: {self.av.keys()}' ) raise excep.BiogemeError(error_msg) if self.av[choice].getValue() == 0.0: return -np.log(0) Vchosen = self.util[choice].getValue() denom = 0.0 for i, V in self.util.items(): if self.av[i].getValue() != 0.0: denom += np.exp(V.getValue() - Vchosen) return -np.log(denom)
def __str__(self): s = self.getClassName() s += f'[choice={self.choice}]' s += 'U=(' first = True for i, e in self.util.items(): if first: s += f'{int(i)}:{e}' first = False else: s += f', {int(i)}:{e}' s += ')' s += 'av=(' first = True for i, e in self.av.items(): if first: s += f'{int(i)}:{e}' first = False else: s += f', {int(i)}:{e}' s += ')' return s
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the signatures of all the children expressions, 2. the name of the expression between < > 3. the id of the expression between { } 4. the number of alternatives between ( ) 5. the id of the expression for the chosen alternative, preceeded by a comma. 6. for each alternative, separated by commas: a. the number of the alternative, as defined by the user, b. the id of the expression for the utility, c. the id of the expression for the availability condition. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ listOfSignatures = [] for e in self.get_children(): listOfSignatures += e.getSignature() signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f'({len(self.util)})' signature += f',{self.choice.get_id()}' for i, e in self.util.items(): signature += f',{i},{e.get_id()},{self.av[i].get_id()}' listOfSignatures += [signature.encode()] return listOfSignatures
class _bioLogLogit(LogLogit): """log of logit formula This expression captures the logarithm of the logit formula. It contains one formula for the target alternative, a dict of formula for the availabilities and a dict of formulas for the utilities It uses only the C++ implementation. """ class _bioLogLogitFullChoiceSet(LogLogit): """This expression captures the logarithm of the logit formula, where all alternatives are supposed to be always available. It contains one formula for the target alternative and a dict of formulas for the utilities. It uses only the C++ implementation. """
[docs]class bioMultSum(Expression): """This expression returns the sum of several other expressions. It is a generalization of 'Plus' for more than two terms """
[docs] def __init__(self, listOfExpressions): """Constructor :param listOfExpressions: list of objects representing the terms of the sum. :type listOfExpressions: list(biogeme.expressions.Expression) :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. :raise BiogemeError: if the list of expressions is empty :raise BiogemeError: if the list of expressions is neither a dict or a list """ Expression.__init__(self) if not listOfExpressions: raise excep.BiogemeError('The argument of bioMultSum cannot be empty') if isinstance(listOfExpressions, dict): for e in listOfExpressions.values(): if isNumeric(e): theExpression = Numeric(e) self.children.append(theExpression) else: if not isinstance(e, Expression): raise excep.BiogemeError(f'This is not a valid expression: {e}') self.children.append(e) elif isinstance(listOfExpressions, list): for e in listOfExpressions: if isNumeric(e): theExpression = Numeric(e) self.children.append(theExpression) else: if not isinstance(e, Expression): raise excep.BiogemeError(f'This is not a valid expression: {e}') self.children.append(e) else: raise excep.BiogemeError('Argument of bioMultSum must be a dict or a list.')
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float """ result = 0.0 for e in self.get_children(): result += e.getValue() return result
def __str__(self): s = 'bioMultSum(' + ', '.join([f'{e}' for e in self.get_children()]) + ')' return s
[docs]class Elem(Expression): """This returns the element of a dictionary. The key is evaluated from an expression and must return an integer, possibly negative. """
[docs] def __init__(self, dictOfExpressions, keyExpression): """Constructor :param dictOfExpressions: dict of objects with numerical keys. :type dictOfExpressions: dict(int: biogeme.expressions.Expression) :param keyExpression: object providing the key of the element to be evaluated. :type keyExpression: biogeme.expressions.Expression :raise BiogemeError: if one of the expressions is invalid, that is neither a numeric value or a biogeme.expressions.Expression object. """ Expression.__init__(self) if isinstance(keyExpression, bool): self.keyExpression = Numeric(1) if keyExpression else Numeric(0) elif isNumeric(keyExpression): self.keyExpression = Numeric(keyExpression) else: if not isinstance(keyExpression, Expression): raise excep.BiogemeError( f'This is not a valid expression: {keyExpression}' ) self.keyExpression = keyExpression #: expression for the key self.children.append(self.keyExpression) self.dictOfExpressions = {} #: dict of expressions for k, v in dictOfExpressions.items(): if isNumeric(v): self.dictOfExpressions[k] = Numeric(v) else: if not isinstance(v, Expression): raise excep.BiogemeError(f'This is not a valid expression: {v}') self.dictOfExpressions[k] = v self.children.append(self.dictOfExpressions[k])
[docs] def getValue(self): """Evaluates the value of the expression :return: value of the expression :rtype: float :raise BiogemeError: if the calcuated key is not present in the dictionary. """ key = int(self.keyExpression.getValue()) if key in self.dictOfExpressions: return self.dictOfExpressions[key].getValue() error_msg = ( f'Key {key} is not present in the dictionary. ' f'Available keys: {self.dictOfExpressions.keys()}' ) raise excep.BiogemeError(error_msg)
def __str__(self): s = '{{' first = True for k, v in self.dictOfExpressions.items(): if first: s += f'{k}:{v}' first = False else: s += f', {k}:{v}' s += f'}}[{self.keyExpression}]' return s
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the signature of the expression defining the key 2. the signatures of all the children expressions, 3. the name of the expression between < > 4. the id of the expression between { } 5. the number of elements between ( ) 6. the id of the expression defining the key 7. for each element: the value of the key and the id of the expression, separated by commas. Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ listOfSignatures = [] listOfSignatures += self.keyExpression.getSignature() for i, e in self.dictOfExpressions.items(): listOfSignatures += e.getSignature() signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f'({len(self.dictOfExpressions)})' signature += f',{self.keyExpression.get_id()}' for i, e in self.dictOfExpressions.items(): signature += f',{i},{e.get_id()}' listOfSignatures += [signature.encode()] return listOfSignatures
[docs]class bioLinearUtility(Expression): """When the utility function is linear, it is expressed as a list of terms, where a parameter multiplies a variable. """
[docs] def __init__(self, listOfTerms): """Constructor :param listOfTerms: a list of tuple. Each tuple contains first a beta parameter, second the name of a variable. :type listOfTerms: list(biogeme.expressions.Expression, biogeme.expressions.Expression) :raises biogeme.exceptions.BiogemeError: if the object is not a list of tuples (parameter, variable) """ Expression.__init__(self) the_error = "" first = True for b, v in listOfTerms: if not isinstance(b, Beta): if first: the_error += ( 'Each element of the bioLinearUtility ' 'must be a tuple (parameter, variable). ' ) first = False the_error += f' Expression {b} is not a parameter.' if not isinstance(v, Variable): if first: the_error += ( 'Each element of the list should be ' 'a tuple (parameter, variable).' ) first = False the_error += f' Expression {v} is not a variable.' if not first: raise excep.BiogemeError(the_error) self.betas, self.variables = zip(*listOfTerms) self.betas = list(self.betas) #: list of parameters self.variables = list(self.variables) #: list of variables self.listOfTerms = list(zip(self.betas, self.variables)) """ List of terms """ self.children += self.betas + self.variables
def __str__(self): return ' + '.join([f'{b} * {x}' for b, x in self.listOfTerms])
[docs] def dict_of_elementary_expression(self, the_type): """Extract a dict with all elementary expressions of a specific type :param the_type: the type of expression :type the_type: TypeOfElementaryExpression :return: returns a dict with the variables appearing in the expression the keys being their names. :rtype: dict(string:biogeme.expressions.Expression) """ if the_type == TypeOfElementaryExpression.BETA: return {x.name: x for x in self.betas} if the_type == TypeOfElementaryExpression.FREE_BETA: return {x.name: x for x in self.betas if x.status == 0} if the_type == TypeOfElementaryExpression.FIXED_BETA: return {x.name: x for x in self.betas if x.status != 0} if the_type == TypeOfElementaryExpression.VARIABLE: return {x.name: x for x in self.variables} return {}
[docs] def getSignature(self): """The signature of a string characterizing an expression. This is designed to be communicated to C++, so that the expression can be reconstructed in this environment. The list contains the following elements: 1. the signatures of all the children expressions, 2. the name of the expression between < > 3. the id of the expression between { } 4. the number of terms in the utility ( ) 5. for each term: a. the id of the beta parameter b. the unique id of the beta parameter c. the name of the parameter d. the id of the variable e. the unique id of the variable f. the name of the variable Consider the following expression: .. math:: 2 \\beta_1 V_1 - \\frac{\\exp(-\\beta_2 V_2) }{ \\beta_3 (\\beta_2 \\geq \\beta_1)}. It is defined as:: 2 * beta1 * Variable1 - expressions.exp(-beta2*Variable2) / (beta3 * (beta2 >= beta1)) And its signature is:: [b'<Numeric>{4780527008},2', b'<Beta>{4780277152}"beta1"[0],0,0', b'<Times>{4780526952}(2),4780527008,4780277152', b'<Variable>{4511837152}"Variable1",5,2', b'<Times>{4780527064}(2),4780526952,4511837152', b'<Beta>{4780277656}"beta2"[0],1,1', b'<UnaryMinus>{4780527120}(1),4780277656', b'<Variable>{4511837712}"Variable2",6,3', b'<Times>{4780527176}(2),4780527120,4511837712', b'<exp>{4780527232}(1),4780527176', b'<Beta>{4780277264}"beta3"[1],2,0', b'<Beta>{4780277656}"beta2"[0],1,1', b'<Beta>{4780277152}"beta1"[0],0,0', b'<GreaterOrEqual>{4780527288}(2),4780277656,4780277152', b'<Times>{4780527344}(2),4780277264,4780527288', b'<Divide>{4780527400}(2),4780527232,4780527344', b'<Minus>{4780527456}(2),4780527064,4780527400'] :return: list of the signatures of an expression and its children. :rtype: list(string) """ listOfSignatures = [] for e in self.get_children(): listOfSignatures += e.getSignature() signature = f'<{self.getClassName()}>' signature += f'{{{self.get_id()}}}' signature += f'({len(self.listOfTerms)})' for b, v in self.listOfTerms: signature += ( f',{b.get_id()},{b.elementaryIndex},{b.name},' f'{v.get_id()},{v.elementaryIndex},{v.name}' ) listOfSignatures += [signature.encode()] return listOfSignatures