######################################## # # @file GRRM.py # @author: Sander van Cranenburgh # @date: 06/11/2015 # ####################################### from biogeme import * from headers import * from loglikelihood import * from statistics import * #Parameters to be estimated # Arguments: # - 1 Name for report; Typically, the same as the variable. # - 2 Starting value. # - 3 Lower bound. # - 4 Upper bound. # - 5 0: estimate the parameter, 1: keep it fixed. # B_FSG = Beta('Floor space groceries',0,-10,10,0) B_FSO = Beta('Floor space other',0,-10,10,0) B_TT = Beta('Travel time',0,-10,10,0) gamma = Beta('gamma',1,0,1,0) FSG2_1 = DefineVariable('FSG2_1', ( FSG2 - FSG1 ) / 1000) FSG3_1 = DefineVariable('FSG3_1', ( FSG3 - FSG1 ) / 1000) FSG4_1 = DefineVariable('FSG4_1', ( FSG4 - FSG1 ) / 1000) FSG5_1 = DefineVariable('FSG5_1', ( FSG5 - FSG1 ) / 1000) FSG1_2 = DefineVariable('FSG1_2', ( FSG1 - FSG2 ) / 1000) FSG3_2 = DefineVariable('FSG3_2', ( FSG3 - FSG2 ) / 1000) FSG4_2 = DefineVariable('FSG4_2', ( FSG4 - FSG2 ) / 1000) FSG5_2 = DefineVariable('FSG5_2', ( FSG5 - FSG2 ) / 1000) FSG1_3 = DefineVariable('FSG1_3', ( FSG1 - FSG3 ) / 1000) FSG2_3 = DefineVariable('FSG2_3', ( FSG2 - FSG3 ) / 1000) FSG4_3 = DefineVariable('FSG4_3', ( FSG4 - FSG3 ) / 1000) FSG5_3 = DefineVariable('FSG5_3', ( FSG5 - FSG3 ) / 1000) FSG1_4 = DefineVariable('FSG1_4', ( FSG1 - FSG4 ) / 1000) FSG2_4 = DefineVariable('FSG2_4', ( FSG2 - FSG4 ) / 1000) FSG3_4 = DefineVariable('FSG3_4', ( FSG3 - FSG4 ) / 1000) FSG5_4 = DefineVariable('FSG5_4', ( FSG5 - FSG4 ) / 1000) FSG1_5 = DefineVariable('FSG1_5', ( FSG1 - FSG5 ) / 1000) FSG2_5 = DefineVariable('FSG2_5', ( FSG2 - FSG5 ) / 1000) FSG3_5 = DefineVariable('FSG3_5', ( FSG3 - FSG5 ) / 1000) FSG4_5 = DefineVariable('FSG4_5', ( FSG4 - FSG5 ) / 1000) # Floorspace Other FSO2_1 = DefineVariable('FSO2_1', ( FSO2 - FSO1 ) / 1000) FSO3_1 = DefineVariable('FSO3_1', ( FSO3 - FSO1 ) / 1000) FSO4_1 = DefineVariable('FSO4_1', ( FSO4 - FSO1 ) / 1000) FSO5_1 = DefineVariable('FSO5_1', ( FSO5 - FSO1 ) / 1000) FSO1_2 = DefineVariable('FSO1_2', ( FSO1 - FSO2 ) / 1000) FSO3_2 = DefineVariable('FSO3_2', ( FSO3 - FSO2 ) / 1000) FSO4_2 = DefineVariable('FSO4_2', ( FSO4 - FSO2 ) / 1000) FSO5_2 = DefineVariable('FSO5_2', ( FSO5 - FSO2 ) / 1000) FSO1_3 = DefineVariable('FSO1_3', ( FSO1 - FSO3 ) / 1000) FSO2_3 = DefineVariable('FSO2_3', ( FSO2 - FSO3 ) / 1000) FSO4_3 = DefineVariable('FSO4_3', ( FSO4 - FSO3 ) / 1000) FSO5_3 = DefineVariable('FSO5_3', ( FSO5 - FSO3 ) / 1000) FSO1_4 = DefineVariable('FSO1_4', ( FSO1 - FSO4 ) / 1000) FSO2_4 = DefineVariable('FSO2_4', ( FSO2 - FSO4 ) / 1000) FSO3_4 = DefineVariable('FSO3_4', ( FSO3 - FSO4 ) / 1000) FSO5_4 = DefineVariable('FSO5_4', ( FSO5 - FSO4 ) / 1000) FSO1_5 = DefineVariable('FSO1_5', ( FSO1 - FSO5 ) / 1000) FSO2_5 = DefineVariable('FSO2_5', ( FSO2 - FSO5 ) / 1000) FSO3_5 = DefineVariable('FSO3_5', ( FSO3 - FSO5 ) / 1000) FSO4_5 = DefineVariable('FSO4_5', ( FSO4 - FSO5 ) / 1000) # Travel time TT2_1 = DefineVariable('TT2_1', ( TT2 - TT1 ) / 100) TT3_1 = DefineVariable('TT3_1', ( TT3 - TT1 ) / 100) TT4_1 = DefineVariable('TT4_1', ( TT4 - TT1 ) / 100) TT5_1 = DefineVariable('TT5_1', ( TT5 - TT1 ) / 100) TT1_2 = DefineVariable('TT1_2', ( TT1 - TT2 ) / 100) TT3_2 = DefineVariable('TT3_2', ( TT3 - TT2 ) / 100) TT4_2 = DefineVariable('TT4_2', ( TT4 - TT2 ) / 100) TT5_2 = DefineVariable('TT5_2', ( TT5 - TT2 ) / 100) TT1_3 = DefineVariable('TT1_3', ( TT1 - TT3 ) / 100) TT2_3 = DefineVariable('TT2_3', ( TT2 - TT3 ) / 100) TT4_3 = DefineVariable('TT4_3', ( TT4 - TT3 ) / 100) TT5_3 = DefineVariable('TT5_3', ( TT5 - TT3 ) / 100) TT1_4 = DefineVariable('TT1_4', ( TT1 - TT4 ) / 100) TT2_4 = DefineVariable('TT2_4', ( TT2 - TT4 ) / 100) TT3_4 = DefineVariable('TT3_4', ( TT3 - TT4 ) / 100) TT5_4 = DefineVariable('TT5_4', ( TT5 - TT4 ) / 100) TT1_5 = DefineVariable('TT1_5', ( TT1 - TT5 ) / 100) TT2_5 = DefineVariable('TT2_5', ( TT2 - TT5 ) / 100) TT3_5 = DefineVariable('TT3_5', ( TT3 - TT5 ) / 100) TT4_5 = DefineVariable('TT4_5', ( TT4 - TT5 ) / 100) # Utility functions R1 = - ( log( gamma + exp( B_FSG * FSG2_1 ) ) + log( gamma + exp( B_FSG * FSG3_1 ) ) + log( gamma + exp( B_FSG * FSG4_1 ) ) + log( gamma + exp( B_FSG * FSG5_1 ) ) + log( gamma + exp( B_FSO * FSO2_1 ) ) + log( gamma + exp( B_FSO * FSO3_1 ) ) + log( gamma + exp( B_FSO * FSO4_1 ) ) + log( gamma + exp( B_FSO * FSO5_1 ) ) + log( gamma + exp( B_TT * TT2_1 ) ) + log( gamma + exp( B_TT * TT3_1 ) ) + log( gamma + exp( B_TT * TT4_1 ) ) + log( gamma + exp( B_TT * TT5_1 ) )) R2 = - ( log( gamma + exp( B_FSG * FSG1_2 ) ) + log( gamma + exp( B_FSG * FSG3_2 ) ) + log( gamma + exp( B_FSG * FSG4_2 ) ) + log( gamma + exp( B_FSG * FSG5_2 ) ) + log( gamma + exp( B_FSO * FSO1_2 ) ) + log( gamma + exp( B_FSO * FSO3_2 ) ) + log( gamma + exp( B_FSO * FSO4_2 ) ) + log( gamma + exp( B_FSO * FSO5_2 ) ) + log( gamma + exp( B_TT * TT1_2 ) ) + log( gamma + exp( B_TT * TT3_2 ) ) + log( gamma + exp( B_TT * TT4_2 ) ) + log( gamma + exp( B_TT * TT5_2 ) )) R3 = - ( log( gamma + exp( B_FSG * FSG1_3 ) ) + log( gamma + exp( B_FSG * FSG2_3 ) ) + log( gamma + exp( B_FSG * FSG4_3 ) ) + log( gamma + exp( B_FSG * FSG5_3 ) ) + log( gamma + exp( B_FSO * FSO1_3 ) ) + log( gamma + exp( B_FSO * FSO2_3 ) ) + log( gamma + exp( B_FSO * FSO4_3 ) ) + log( gamma + exp( B_FSO * FSO5_3 ) ) + log( gamma + exp( B_TT * TT1_3 ) ) + log( gamma + exp( B_TT * TT2_3 ) ) + log( gamma + exp( B_TT * TT4_3 ) ) + log( gamma + exp( B_TT * TT5_3 ) )) R4 = - ( log( gamma + exp( B_FSG * FSG1_4 ) ) + log( gamma + exp( B_FSG * FSG2_4 ) ) + log( gamma + exp( B_FSG * FSG3_4 ) ) + log( gamma + exp( B_FSG * FSG5_4 ) ) + log( gamma + exp( B_FSO * FSO1_4 ) ) + log( gamma + exp( B_FSO * FSO2_4 ) ) + log( gamma + exp( B_FSO * FSO3_4 ) ) + log( gamma + exp( B_FSO * FSO5_4 ) ) + log( gamma + exp( B_TT * TT1_4 ) ) + log( gamma + exp( B_TT * TT2_4 ) ) + log( gamma + exp( B_TT * TT3_4 ) ) + log( gamma + exp( B_TT * TT5_4 ) )) R5 = - ( log( gamma + exp( B_FSG * FSG1_5 ) ) + log( gamma + exp( B_FSG * FSG2_5 ) ) + log( gamma + exp( B_FSG * FSG3_5 ) ) + log( gamma + exp( B_FSG * FSG4_5 ) ) + log( gamma + exp( B_FSO * FSO1_5 ) ) + log( gamma + exp( B_FSO * FSO2_5 ) ) + log( gamma + exp( B_FSO * FSO3_5 ) ) + log( gamma + exp( B_FSO * FSO4_5 ) ) + log( gamma + exp( B_TT * TT1_5 ) ) + log( gamma + exp( B_TT * TT2_5 ) ) + log( gamma + exp( B_TT * TT3_5 ) ) + log( gamma + exp( B_TT * TT4_5 ) ) ) # Associate utility functions with the numbering of alternatives V = {1: R1, 2: R2, 3: R3, 4: R4, 5: R5} # Associate the availability conditions with the alternatives one = DefineVariable('one',1) av = {1: one, 2: one, 3: one, 4: one, 5: one} # The choice model is a logit, with availability conditions prob = bioLogit(V,av,CHOICE) logP = log(prob) # Defines an itertor on the data rowIterator('obsIter') # DEfine the likelihood function for the estimation BIOGEME_OBJECT.ESTIMATE = Sum(logP,'obsIter') # Statistics nullLoglikelihood(av,'obsIter') choiceSet = [1,2,3,4,5] cteLoglikelihood(choiceSet,CHOICE,'obsIter') availabilityStatistics(av,'obsIter') BIOGEME_OBJECT.PARAMETERS['optimizationAlgorithm'] = "CFSQP" BIOGEME_OBJECT.PARAMETERS['numberOfThreads'] = "2"