Important: Read this before posting to this forum

  1. This forum is for questions related to the use of Apollo. We will answer some general choice modelling questions too, where appropriate, and time permitting. We cannot answer questions about how to estimate choice models with other software packages.
  2. There is a very detailed manual for Apollo available at http://www.ApolloChoiceModelling.com/manual.html. This contains detailed descriptions of the various Apollo functions, and numerous examples are available at http://www.ApolloChoiceModelling.com/examples.html. In addition, help files are available for all functions, using e.g. ?apollo_mnl
  3. Before asking a question on the forum, users are kindly requested to follow these steps:
    1. Check that the same issue has not already been addressed in the forum - there is a search tool.
    2. Ensure that the correct syntax has been used. For any function, detailed instructions are available directly in Apollo, e.g. by using ?apollo_mnl for apollo_mnl
    3. Check the frequently asked questions section on the Apollo website, which discusses some common issues/failures. Please see http://www.apollochoicemodelling.com/faq.html
    4. Make sure that R is using the latest official release of Apollo.
  4. If the above steps do not resolve the issue, then users should follow these steps when posting a question:
    1. provide full details on the issue, including the entire code and output, including any error messages
    2. posts will not immediately appear on the forum, but will be checked by a moderator first. We check the forum at least twice a week. It may thus take a couple of days for your post to appear and before we reply. There is no need to submit the post multiple times.

speedTest function in multiple cores

Ask general questions about model specification and estimation that are not Apollo specific but relevant to Apollo users.
Post Reply
tue_qiang
Posts: 9
Joined: 02 May 2023, 17:13

speedTest function in multiple cores

Post by tue_qiang »

I established a mixed logit model and about 20 random parameters are set. when I use 1000 MLHS draws for these parameter, the model runs a very long time, about several days. therefore, I use the speedTest function to explore the most suitable number of cores and the seconds used in likelihood calculation keep little changed since the 3 cores, no matter the draws are 500, 1000. Does it mean 3 cores is are the most suitable cores' number? I do not understand it, because I have so many random parameters and draws and in my mind, more cores lead higher model estimation speed.
stephanehess
Site Admin
Posts: 1235
Joined: 24 Apr 2020, 16:29

Re: speedTest function in multiple cores

Post by stephanehess »

Hi

yes, this suggests that 3 cores is a good number to use. The benefits of multiple cores might decrease if you use too many cores

Stephane
--------------------------------
Stephane Hess
www.stephanehess.me.uk
tue_qiang
Posts: 9
Joined: 02 May 2023, 17:13

Re: speedTest function in multiple cores

Post by tue_qiang »

Hi Stephane,

the running time is so long now, if i want to speed up it, what method could help it? i can apply for using supercomputer of my university, but I do not know if it will work.

Best,
Qiang
stephanehess
Site Admin
Posts: 1235
Joined: 24 Apr 2020, 16:29

Re: speedTest function in multiple cores

Post by stephanehess »

yes, a faster computer will help
--------------------------------
Stephane Hess
www.stephanehess.me.uk
janak12_jp
Posts: 15
Joined: 08 Sep 2021, 16:52

Re: speedTest function in multiple cores

Post by janak12_jp »

Hi Prof. Hess,

I am facing a similar issue. I am computing latent-class latent-variable model. It has 7 random parameters (all LVs) and a total of 24 OL models for LVs. It is running since last 20 days on my 8 core PC (I am using 7 cores for computation). I have around 7200 observations for 702 IDs. Can you suggest how to increase the speed (other than using HPC, of course that is one of the solutions)?
My code is as below:

Code: Select all

## Clear memory

rm(list = ls())

## Load apollo library

library(apollo)

## Initialise code

apollo_initialise()

## Set core controls

apollo_control = list(
  modelName = "LC with MMNL",
  modelDescr = "LC with random coefficients",
  indivID = "ID",
  nCores = 7,
  outputDirectory = "output"
)

## Load data

database = read.csv("Analysis.csv", header = T)

## Define model parameters
## Vector of parameters to be estimated

apollo_beta = c(
  ## class-specific choice utility 
  # class a
  asc_car_a = 0, car_ivtt_a = 0, car_egt_a = 0, car_tc_a = 0, car_pc_a = 0, lm_innv_car_a = 0, lm_skpt_car_a = 0, lm_risk_car_a = 0, lm_use_car_a = 0, lm_sn_car_a = 0, lm_env_car_a = 0, 
  asc_pt_a = 0, pt_ivtt_a = 0, pt_act_a = 0, pt_egt_a = 0, pt_wt_a = 0, pt_tc_a = 0, lm_innv_pt_a = 0, lm_skpt_pt_a = 0, lm_risk_pt_a = 0, lm_use_pt_a = 0, lm_sn_pt_a = 0, lm_env_pt_a = 0, 
  asc_ecar_a = 0, ecar_ivtt_a = 0, ecar_act_a = 0, ecar_egt_a = 0, ecar_tc_a = 0, ecar_av_a = 0, ecar_wt_a = 0, lm_innv_ecar_a = 0, lm_skpt_ecar_a = 0, lm_risk_ecar_a = 0, lm_use_ecar_a = 0, lm_sn_ecar_a = 0, lm_env_ecar_a = 0, 
  asc_ebike_a = 0, ebike_ivtt_a = 0, ebike_act_a = 0, ebike_egt_a = 0, ebike_tc_a = 0, ebike_av_a = 0, ebike_wt_a = 0, lm_innv_ebike_a = 0, lm_skpt_ebike_a = 0, lm_risk_ebike_a = 0, lm_use_ebike_a = 0, lm_sn_ebike_a = 0, lm_env_ebike_a = 0, 
  asc_escoot_a = 0, escoot_ivtt_a = 0, escoot_act_a = 0, escoot_egt_a = 0, escoot_tc_a = 0, escoot_av_a = 0, escoot_wt_a = 0, lm_innv_escoot_a = 0, lm_skpt_escoot_a = 0, lm_risk_escoot_a = 0, lm_use_escoot_a = 0, lm_sn_escoot_a = 0, lm_env_escoot_a = 0, 
  asc_aebike_a = 0, aebike_ivtt_a = 0, aebike_wk_a = 0, aebike_tc_a = 0, aebike_av_a = 0, aebike_wt_a = 0, lm_innv_aebike_a = 0, lm_skpt_aebike_a = 0, lm_risk_aebike_a = 0, lm_use_aebike_a = 0, lm_sn_aebike_a = 0, lm_env_aebike_a = 0, 
  asc_aescoot_a = 0, aescoot_ivtt_a = 0, aescoot_wk_a = 0, aescoot_tc_a = 0, aescoot_av_a = 0, aescoot_wt_a = 0, lm_innv_aescoot_a = 0, lm_skpt_aescoot_a = 0, lm_risk_aescoot_a = 0, lm_use_aescoot_a = 0, lm_sn_aescoot_a = 0, lm_env_aescoot_a = 0, 
  asc_awalk_a = 0, awalk_a = 0, lm_risk_awalk_a = 0, lm_use_awalk_a = 0, 
  
  # class b
  asc_car_b = 0, car_ivtt_b = 0, car_egt_b = 0, car_tc_b = 0, car_pc_b = 0, lm_innv_car_b = 0, lm_skpt_car_b = 0, lm_risk_car_b = 0, lm_use_car_b = 0, lm_sn_car_b = 0, lm_env_car_b = 0, 
  asc_pt_b = 0, pt_ivtt_b = 0, pt_act_b = 0, pt_egt_b = 0, pt_wt_b = 0, pt_tc_b = 0, lm_innv_pt_b = 0, lm_skpt_pt_b = 0, lm_risk_pt_b = 0, lm_use_pt_b = 0, lm_sn_pt_b = 0, lm_env_pt_b = 0,
  asc_ecar_b = 0, ecar_ivtt_b = 0, ecar_act_b = 0, ecar_egt_b = 0, ecar_tc_b = 0, ecar_av_b = 0, ecar_wt_b = 0, lm_innv_ecar_b = 0, lm_skpt_ecar_b = 0, lm_risk_ecar_b = 0, lm_use_ecar_b = 0, lm_sn_ecar_b = 0, lm_env_ecar_b = 0, 
  asc_ebike_b = 0, ebike_ivtt_b = 0, ebike_act_b = 0, ebike_egt_b = 0, ebike_tc_b = 0, ebike_av_b = 0, ebike_wt_b = 0, lm_innv_ebike_b = 0, lm_skpt_ebike_b = 0, lm_risk_ebike_b = 0, lm_use_ebike_b = 0, lm_sn_ebike_b = 0, lm_env_ebike_b = 0, 
  asc_escoot_b = 0, escoot_ivtt_b = 0, escoot_act_b = 0, escoot_egt_b = 0, escoot_tc_b = 0, escoot_av_b = 0, escoot_wt_b = 0, lm_innv_escoot_b = 0, lm_skpt_escoot_b = 0, lm_risk_escoot_b = 0, lm_use_escoot_b = 0, lm_sn_escoot_b = 0, lm_env_escoot_b = 0, 
  asc_aebike_b = 0, aebike_ivtt_b = 0, aebike_wk_b = 0, aebike_tc_b = 0, aebike_av_b = 0, aebike_wt_b = 0, lm_innv_aebike_b = 0, lm_skpt_aebike_b = 0, lm_risk_aebike_b = 0, lm_use_aebike_b = 0, lm_sn_aebike_b = 0, lm_env_aebike_b = 0, 
  asc_aescoot_b = 0, aescoot_ivtt_b = 0, aescoot_wk_b = 0, aescoot_tc_b = 0, aescoot_av_b = 0, aescoot_wt_b = 0, lm_innv_aescoot_b = 0, lm_skpt_aescoot_b = 0, lm_risk_aescoot_b = 0, lm_use_aescoot_b = 0, lm_sn_aescoot_b = 0, lm_env_aescoot_b = 0, 
  asc_awalk_b = 0, awalk_b = 0, lm_risk_awalk_b = 0, lm_use_awalk_b = 0,
  
  # class c
  asc_car_c = 0, car_ivtt_c = 0, car_egt_c = 0, car_tc_c = 0, car_pc_c = 0, lm_innv_car_c = 0, lm_skpt_car_c = 0, lm_risk_car_c = 0, lm_use_car_c = 0, lm_sn_car_c = 0, lm_env_car_c = 0, 
  asc_pt_c = 0, pt_ivtt_c = 0, pt_act_c = 0, pt_egt_c = 0, pt_wt_c = 0, pt_tc_c = 0, lm_innv_pt_c = 0, lm_skpt_pt_c = 0, lm_risk_pt_c = 0, lm_use_pt_c = 0, lm_sn_pt_c = 0, lm_env_pt_c = 0, 
  asc_ecar_c = 0, ecar_ivtt_c = 0, ecar_act_c = 0, ecar_egt_c = 0, ecar_tc_c = 0, ecar_av_c = 0, ecar_wt_c = 0, lm_innv_ecar_c = 0, lm_skpt_ecar_c = 0, lm_risk_ecar_c = 0, lm_use_ecar_c = 0, lm_sn_ecar_c = 0, lm_env_ecar_c = 0, 
  asc_ebike_c = 0, ebike_ivtt_c = 0, ebike_act_c = 0, ebike_egt_c = 0, ebike_tc_c = 0, ebike_av_c = 0, ebike_wt_c = 0, lm_innv_ebike_c = 0, lm_skpt_ebike_c = 0, lm_risk_ebike_c = 0, lm_use_ebike_c = 0, lm_sn_ebike_c = 0, lm_env_ebike_c = 0, 
  asc_escoot_c = 0, escoot_ivtt_c = 0, escoot_act_c = 0, escoot_egt_c = 0, escoot_tc_c = 0, escoot_av_c = 0, escoot_wt_c = 0, lm_innv_escoot_c = 0, lm_skpt_escoot_c = 0, lm_risk_escoot_c = 0, lm_use_escoot_c = 0, lm_sn_escoot_c = 0, lm_env_escoot_c = 0, 
  asc_aebike_c = 0, aebike_ivtt_c = 0, aebike_wk_c = 0, aebike_tc_c = 0, aebike_av_c = 0, aebike_wt_c = 0, lm_innv_aebike_c = 0, lm_skpt_aebike_c = 0, lm_risk_aebike_c = 0, lm_use_aebike_c = 0, lm_sn_aebike_c = 0, lm_env_aebike_c = 0, 
  asc_aescoot_c = 0, aescoot_ivtt_c = 0, aescoot_wk_c = 0, aescoot_tc_c = 0, aescoot_av_c = 0, aescoot_wt_c = 0, lm_innv_aescoot_c = 0, lm_skpt_aescoot_c = 0, lm_risk_aescoot_c = 0, lm_use_aescoot_c = 0, lm_sn_aescoot_c = 0, lm_env_aescoot_c = 0, 
  asc_awalk_c = 0, awalk_c = 0, lm_risk_awalk_c = 0, lm_use_awalk_c = 0,
  
  ## class-membership model
  # class a
  delta_a = 0, gm_female_a = 0, gm_age1_a = 0, gm_age2_a = 0, gm_age3_a = 0, gm_hhinc1_a = 0, gm_hhinc2_a = 0, gm_hhinc3_a = 0,
  gm_ncar0_a = 0, gm_ncar1_a = 0, gm_ncar2_a = 0, gm_nbic0_a = 0, gm_nbic1_a = 0, gm_nbic2_a = 0, gm_hhC_a = 0, lm_hab_a = 0,
  
  # class b
  delta_b = 0, gm_female_b = 0, gm_age1_b = 0, gm_age2_b = 0, gm_age3_b = 0, gm_hhinc1_b = 0, gm_hhinc2_b = 0, gm_hhinc3_b = 0,
  gm_ncar0_b = 0, gm_ncar1_b = 0, gm_ncar2_b = 0, gm_nbic0_b = 0, gm_nbic1_b = 0, gm_nbic2_b = 0, gm_hhC_b = 0, lm_hab_b = 0,
  
  # class c
  delta_c = 0, gm_female_c = 0, gm_age1_c = 0, gm_age2_c = 0, gm_age3_c = 0, gm_hhinc1_c = 0, gm_hhinc2_c = 0, gm_hhinc3_c = 0,
  gm_ncar0_c = 0, gm_ncar1_c = 0, gm_ncar2_c = 0, gm_nbic0_c = 0, gm_nbic1_c = 0, gm_nbic2_c = 0, gm_hhC_c = 0, lm_hab_c = 0,
  
  ## structural model for LVs
  gamma_age1 = 0, gamma_age2 = 0, gamma_age3 = 0, gamma_hhinc1 = 0, gamma_hhinc2 = 0, gamma_hhinc3 = 0, gamma_grad = 0, gamma_female = 0, gamma_ncar0 = 0, gamma_ncar1 = 0, gamma_ncar2 = 0,
  gamma_nbic0 = 0, gamma_nbic1 = 0, gamma_nbic2 = 0,
  
  ## measurement model - LV to indicators
  zeta_innv1 = 0, zeta_innv2 = 0, zeta_innv3 = 0, zeta_skpt1 = 0, zeta_skpt2 = 0, zeta_skpt3 = 0, zeta_risk1 = 0, zeta_risk2 = 0, zeta_risk3 = 0,  
  zeta_use1 = 0, zeta_use2 = 0, zeta_use3 = 0, zeta_use4 = 0, zeta_use5 = 0, zeta_sn1 = 0, zeta_sn2 = 0, 
  zeta_env1 = 0, zeta_env2 = 0, zeta_env3 = 0, zeta_hab1 = 0, zeta_hab2 = 0, zeta_hab3 = 0, zeta_hab4 = 0, zeta_hab5 = 0,
  
  tau_innv1_1 = -2, tau_innv1_2 = -1, tau_innv1_3 = 1, tau_innv1_4 = 2,
  tau_innv2_1 = -2, tau_innv2_2 = -1, tau_innv2_3 = 1, tau_innv2_4 = 2,
  tau_innv3_1 = -2, tau_innv3_2 = -1, tau_innv3_3 = 1, tau_innv3_4 = 2,
  tau_skpt1_1 = -2, tau_skpt1_2 = -1, tau_skpt1_3 = 1, tau_skpt1_4 = 2,
  tau_skpt2_1 = -2, tau_skpt2_2 = -1, tau_skpt2_3 = 1, tau_skpt2_4 = 2,
  tau_skpt3_1 = -2, tau_skpt3_2 = -1, tau_skpt3_3 = 1, tau_skpt3_4 = 2,
  tau_risk1_1 = -2, tau_risk1_2 = -1, tau_risk1_3 = 1, tau_risk1_4 = 2,
  tau_risk2_1 = -2, tau_risk2_2 = -1, tau_risk2_3 = 1, tau_risk2_4 = 2,
  tau_risk3_1 = -2, tau_risk3_2 = -1, tau_risk3_3 = 1, tau_risk3_4 = 2,
  tau_use1_1 = -2, tau_use1_2 = -1, tau_use1_3 = 1, tau_use1_4 = 2,
  tau_use2_1 = -2, tau_use2_2 = -1, tau_use2_3 = 1, tau_use2_4 = 2,
  tau_use3_1 = -2, tau_use3_2 = -1, tau_use3_3 = 1, tau_use3_4 = 2,
  tau_use4_1 = -2, tau_use4_2 = -1, tau_use4_3 = 1, tau_use4_4 = 2,
  tau_use5_1 = -2, tau_use5_2 = -1, tau_use5_3 = 1, tau_use5_4 = 2,
  tau_sn1_1 = -2, tau_sn1_2 = -1, tau_sn1_3 = 1, tau_sn1_4 = 2,
  tau_sn2_1 = -2, tau_sn2_2 = -1, tau_sn2_3 = 1, tau_sn2_4 = 2,
  tau_env1_1 = -2, tau_env1_2 = -1, tau_env1_3 = 1, tau_env1_4 = 2,
  tau_env2_1 = -2, tau_env2_2 = -1, tau_env2_3 = 1, tau_env2_4 = 2,
  tau_env3_1 = -2, tau_env3_2 = -1, tau_env3_3 = 1, tau_env3_4 = 2,
  tau_hab1_1 = -2, tau_hab1_2 = -1, tau_hab1_3 = 1, tau_hab1_4 = 2,
  tau_hab2_1 = -2, tau_hab2_2 = -1, tau_hab2_3 = 1, tau_hab2_4 = 2,
  tau_hab3_1 = -2, tau_hab3_2 = -1, tau_hab3_3 = 1, tau_hab3_4 = 2,
  tau_hab4_1 = -2, tau_hab4_2 = -1, tau_hab4_3 = 1, tau_hab4_4 = 2,
  tau_hab5_1 = -2, tau_hab5_2 = -1, tau_hab5_3 = 1, tau_hab5_4 = 2
)

## Define fixed parameters in model

apollo_fixed = c("asc_car_a", "asc_awalk_a", "asc_car_b", "asc_awalk_b", "asc_car_c", "asc_awalk_c", "delta_c", "gm_female_c", "gm_age1_c", "gm_age2_c", "gm_age3_c", "gm_hhinc1_c", 
                 "gm_hhinc2_c", "gm_hhinc3_c", "gm_ncar0_c", "gm_ncar1_c", "gm_ncar2_c", "gm_nbic0_c", "gm_nbic1_c", "gm_nbic2_c", "gm_hhC_c", "lm_hab_c",
                 "gm_age3_a", "gm_age3_b", "gm_hhinc3_a", "gm_hhinc3_b", "gm_ncar0_a", "gm_ncar0_b", "gm_nbic0_a", "gm_nbic0_b",
                 "gamma_age3", "gamma_hhinc3", "gamma_ncar0", "gamma_nbic0")

## Define random parameters

apollo_draws = list(
  interDrawsType = "mlhs",
  interNDraws = 1000,
  interNormDraws = c("eta1", "eta2", "eta3", "eta4", "eta5", "eta6", "eta7")
)

apollo_randCoeff = function(apollo_beta, apollo_inputs) {
  randcoeff = list()
  
  randcoeff[["innv"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + eta1
  randcoeff[["skpt"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + eta2
  randcoeff[["risk"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + eta3
  randcoeff[["use"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + eta4
  randcoeff[["sn"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + eta5
  randcoeff[["env"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + eta6
  randcoeff[["hab"]] = gamma_female*female + gamma_age1*(age==1) + gamma_age2*(age==3) + gamma_age3*(age==5) + gamma_hhinc1*(hhinc==1) + gamma_hhinc2*(hhinc==3) + gamma_hhinc3*(hhinc==5) + gamma_grad*edu + 
    gamma_ncar0*(ncar==0) + gamma_ncar1*(ncar==1) + gamma_ncar2*(ncar==2) + gamma_nbic0*(nbic==0) + gamma_nbic1*(nbic==1) + gamma_nbic2*(nbic==2) + eta7
  
  return(randcoeff)
}

## Define latent class components 

apollo_lcPars = function(apollo_beta, apollo_inputs){
  lcpars = list()
  
  lcpars[["asc_car"]] = list(asc_car_a, asc_car_b, asc_car_c)
  lcpars[["asc_pt"]] = list(asc_pt_a, asc_pt_b, asc_pt_c)
  lcpars[["asc_ecar"]] = list(asc_ecar_a, asc_ecar_b, asc_ecar_c)
  lcpars[["asc_ebike"]] = list(asc_ebike_a, asc_ebike_b, asc_ebike_c)
  lcpars[["asc_escoot"]] = list(asc_escoot_a, asc_escoot_b, asc_escoot_c)
  lcpars[["asc_aebike"]] = list(asc_aebike_a, asc_aebike_b, asc_aebike_c)
  lcpars[["asc_aescoot"]] = list(asc_aescoot_a, asc_aescoot_b, asc_aescoot_c)
  lcpars[["asc_awalk"]] = list(asc_awalk_a, asc_awalk_b, asc_awalk_c)
  lcpars[["b_car_ivtt"]] = list(car_ivtt_a, car_ivtt_b, car_ivtt_c)
  lcpars[["b_car_egt"]] = list(car_egt_a, car_egt_b, car_egt_c)
  lcpars[["b_car_tc"]] = list(car_tc_a, car_tc_b, car_tc_c)
  lcpars[["b_car_pc"]] = list(car_pc_a, car_pc_b, car_pc_c)
  lcpars[["b_pt_ivtt"]] = list(pt_ivtt_a, pt_ivtt_b, pt_ivtt_c)
  lcpars[["b_pt_act"]] = list(pt_act_a, pt_act_b, pt_act_c)
  lcpars[["b_pt_egt"]] = list(pt_egt_a, pt_egt_b, pt_egt_c)
  lcpars[["b_pt_wt"]] = list(pt_wt_a, pt_wt_b, pt_wt_c)
  lcpars[["b_pt_tc"]] = list(pt_tc_a, pt_tc_b, pt_tc_c)
  lcpars[["b_ecar_ivtt"]] = list(ecar_ivtt_a, ecar_ivtt_b, ecar_ivtt_c)
  lcpars[["b_ecar_act"]] = list(ecar_act_a, ecar_act_b, ecar_act_c)
  lcpars[["b_ecar_egt"]] = list(ecar_egt_a, ecar_egt_b, ecar_egt_c)
  lcpars[["b_ecar_tc"]] = list(ecar_tc_a, ecar_tc_b, ecar_tc_c)
  lcpars[["b_ecar_av"]] = list(ecar_av_a, ecar_av_b, ecar_av_c)
  lcpars[["b_ecar_wt"]] = list(ecar_wt_a, ecar_wt_b, ecar_wt_c)
  lcpars[["b_ebike_ivtt"]] = list(ebike_ivtt_a, ebike_ivtt_b, ebike_ivtt_c)
  lcpars[["b_ebike_act"]] = list(ebike_act_a, ebike_act_b, ebike_act_c)
  lcpars[["b_ebike_egt"]] = list(ebike_egt_a, ebike_egt_b, ebike_egt_c)
  lcpars[["b_ebike_tc"]] = list(ebike_tc_a, ebike_tc_b, ebike_tc_c)
  lcpars[["b_ebike_av"]] = list(ebike_av_a, ebike_av_b, ebike_av_c)
  lcpars[["b_ebike_wt"]] = list(ebike_wt_a, ebike_wt_b, ebike_wt_c)
  lcpars[["b_escoot_ivtt"]] = list(escoot_ivtt_a, escoot_ivtt_b, escoot_ivtt_c)
  lcpars[["b_escoot_act"]] = list(escoot_act_a, escoot_act_b, escoot_act_c)
  lcpars[["b_escoot_egt"]] = list(escoot_egt_a, escoot_egt_b, escoot_egt_c)
  lcpars[["b_escoot_tc"]] = list(escoot_tc_a, escoot_tc_b, escoot_tc_c)
  lcpars[["b_escoot_av"]] = list(escoot_av_a, escoot_av_b, escoot_av_c)
  lcpars[["b_escoot_wt"]] = list(escoot_wt_a, escoot_wt_b, escoot_wt_c)
  lcpars[["b_aebike_ivtt"]] = list(aebike_ivtt_a, aebike_ivtt_b, aebike_ivtt_c)
  lcpars[["b_aebike_wk"]] = list(aebike_wk_a, aebike_wk_b, aebike_wk_c)
  lcpars[["b_aebike_tc"]] = list(aebike_tc_a, aebike_tc_b, aebike_tc_c)
  lcpars[["b_aebike_av"]] = list(aebike_av_a, aebike_av_b, aebike_av_c)
  lcpars[["b_aebike_wt"]] = list(aebike_wt_a, aebike_wt_b, aebike_wt_c)
  lcpars[["b_aescoot_ivtt"]] = list(aescoot_ivtt_a, aescoot_ivtt_b, aescoot_ivtt_c)
  lcpars[["b_aescoot_wk"]] = list(aescoot_wk_a, aescoot_wk_b, aescoot_wk_c)
  lcpars[["b_aescoot_tc"]] = list(aescoot_tc_a, aescoot_tc_b, aescoot_tc_c)
  lcpars[["b_aescoot_av"]] = list(aescoot_av_a, aescoot_av_b, aescoot_av_c)
  lcpars[["b_aescoot_wt"]] = list(aescoot_wt_a, aescoot_wt_b, aescoot_wt_c)
  lcpars[["b_awalk"]] = list(awalk_a, awalk_b, awalk_c)
  lcpars[["lm_innv_car"]] = list(lm_innv_car_a, lm_innv_car_b, lm_innv_car_c)
  lcpars[["lm_skpt_car"]] = list(lm_skpt_car_a, lm_skpt_car_b, lm_skpt_car_c)
  lcpars[["lm_risk_car"]] = list(lm_risk_car_a, lm_risk_car_b, lm_risk_car_c)
  lcpars[["lm_use_car"]] = list(lm_use_car_a, lm_use_car_b, lm_use_car_c)
  lcpars[["lm_sn_car"]] = list(lm_sn_car_a, lm_sn_car_b, lm_sn_car_c)
  lcpars[["lm_env_car"]] = list(lm_env_car_a, lm_env_car_b, lm_env_car_c)
  lcpars[["lm_innv_pt"]] = list(lm_innv_pt_a, lm_innv_pt_b, lm_innv_pt_c)
  lcpars[["lm_skpt_pt"]] = list(lm_skpt_pt_a, lm_skpt_pt_b, lm_skpt_pt_c)
  lcpars[["lm_risk_pt"]] = list(lm_risk_pt_a, lm_risk_pt_b, lm_risk_pt_c)
  lcpars[["lm_use_pt"]] = list(lm_use_pt_a, lm_use_pt_b, lm_use_pt_c)
  lcpars[["lm_sn_pt"]] = list(lm_sn_pt_a, lm_sn_pt_b, lm_sn_pt_c)
  lcpars[["lm_env_pt"]] = list(lm_env_pt_a, lm_env_pt_b, lm_env_pt_c)
  lcpars[["lm_innv_ecar"]] = list(lm_innv_ecar_a, lm_innv_ecar_b, lm_innv_ecar_c)
  lcpars[["lm_skpt_ecar"]] = list(lm_skpt_ecar_a, lm_skpt_ecar_b, lm_skpt_ecar_c)
  lcpars[["lm_risk_ecar"]] = list(lm_risk_ecar_a, lm_risk_ecar_b, lm_risk_ecar_c)
  lcpars[["lm_use_ecar"]] = list(lm_use_ecar_a, lm_use_ecar_b, lm_use_ecar_c)
  lcpars[["lm_sn_ecar"]] = list(lm_sn_ecar_a, lm_sn_ecar_b, lm_sn_ecar_c)
  lcpars[["lm_env_ecar"]] = list(lm_env_ecar_a, lm_env_ecar_b, lm_env_ecar_c)
  lcpars[["lm_innv_ebike"]] = list(lm_innv_ebike_a, lm_innv_ebike_b, lm_innv_ebike_c)
  lcpars[["lm_skpt_ebike"]] = list(lm_skpt_ebike_a, lm_skpt_ebike_b, lm_skpt_ebike_c)
  lcpars[["lm_risk_ebike"]] = list(lm_risk_ebike_a, lm_risk_ebike_b, lm_risk_ebike_c)
  lcpars[["lm_use_ebike"]] = list(lm_use_ebike_a, lm_use_ebike_b, lm_use_ebike_c)
  lcpars[["lm_sn_ebike"]] = list(lm_sn_ebike_a, lm_sn_ebike_b, lm_sn_ebike_c)
  lcpars[["lm_env_ebike"]] = list(lm_env_ebike_a, lm_env_ebike_b, lm_env_ebike_c)
  lcpars[["lm_innv_escoot"]] = list(lm_innv_escoot_a, lm_innv_escoot_b, lm_innv_escoot_c)
  lcpars[["lm_skpt_escoot"]] = list(lm_skpt_escoot_a, lm_skpt_escoot_b, lm_skpt_escoot_c)
  lcpars[["lm_risk_escoot"]] = list(lm_risk_escoot_a, lm_risk_escoot_b, lm_risk_escoot_c)
  lcpars[["lm_use_escoot"]] = list(lm_use_escoot_a, lm_use_escoot_b, lm_use_escoot_c)
  lcpars[["lm_sn_escoot"]] = list(lm_sn_escoot_a, lm_sn_escoot_b, lm_sn_escoot_c)
  lcpars[["lm_env_escoot"]] = list(lm_env_escoot_a, lm_env_escoot_b, lm_env_escoot_c)
  lcpars[["lm_innv_aebike"]] = list(lm_innv_aebike_a, lm_innv_aebike_b, lm_innv_aebike_c)
  lcpars[["lm_skpt_aebike"]] = list(lm_skpt_aebike_a, lm_skpt_aebike_b, lm_skpt_aebike_c)
  lcpars[["lm_risk_aebike"]] = list(lm_risk_aebike_a, lm_risk_aebike_b, lm_risk_aebike_c)
  lcpars[["lm_use_aebike"]] = list(lm_use_aebike_a, lm_use_aebike_b, lm_use_aebike_c)
  lcpars[["lm_sn_aebike"]] = list(lm_sn_aebike_a, lm_sn_aebike_b, lm_sn_aebike_c)
  lcpars[["lm_env_aebike"]] = list(lm_env_aebike_a, lm_env_aebike_b, lm_env_aebike_c)
  lcpars[["lm_innv_aescoot"]] = list(lm_innv_aescoot_a, lm_innv_aescoot_b, lm_innv_aescoot_c)
  lcpars[["lm_skpt_aescoot"]] = list(lm_skpt_aescoot_a, lm_skpt_aescoot_b, lm_skpt_aescoot_c)
  lcpars[["lm_risk_aescoot"]] = list(lm_risk_aescoot_a, lm_risk_aescoot_b, lm_risk_aescoot_c)
  lcpars[["lm_use_aescoot"]] = list(lm_use_aescoot_a, lm_use_aescoot_b, lm_use_aescoot_c)
  lcpars[["lm_sn_aescoot"]] = list(lm_sn_aescoot_a, lm_sn_aescoot_b, lm_sn_aescoot_c)
  lcpars[["lm_env_aescoot"]] = list(lm_env_aescoot_a, lm_env_aescoot_b, lm_env_aescoot_c)
  lcpars[["lm_risk_awalk"]] = list(lm_risk_awalk_a, lm_risk_awalk_b, lm_risk_awalk_c)
  lcpars[["lm_use_awalk"]] = list(lm_use_awalk_a, lm_use_awalk_b, lm_use_awalk_c)
  
  ## Class segmentation model 
  V = list()
  
  V[["class_a"]] = delta_a + gm_age1_a*(age==1) + gm_age2_a*(age==3) + gm_age3_a*(age==5) + gm_female_a*female + gm_hhinc1_a*(hhinc==1) + gm_hhinc2_a*(hhinc==3) + gm_hhinc3_a*(hhinc==5) + 
    gm_ncar0_a*(ncar==0) + gm_ncar1_a*(ncar==1) + gm_ncar2_a*(ncar==2) + gm_nbic0_a*(nbic==0) + gm_nbic1_a*(nbic==1) + gm_nbic2_a*(nbic==2) + gm_hhC_a*hhC + lm_hab_a*hab
  V[["class_b"]] = delta_b + gm_age1_b*(age==1) + gm_age2_b*(age==3) + gm_age3_b*(age==5) + gm_female_b*female + gm_hhinc1_b*(hhinc==1) + gm_hhinc2_b*(hhinc==3) + gm_hhinc3_b*(hhinc==5) + 
    gm_ncar0_b*(ncar==0) + gm_ncar1_b*(ncar==1) + gm_ncar2_b*(ncar==2) + gm_nbic0_b*(nbic==0) + gm_nbic1_b*(nbic==1) + gm_nbic2_b*(nbic==2) + gm_hhC_b*hhC + lm_hab_b*hab
  V[["class_c"]] = delta_c + gm_age1_c*(age==1) + gm_age2_c*(age==3) + gm_age3_c*(age==5) + gm_female_c*female + gm_hhinc1_c*(hhinc==1) + gm_hhinc2_c*(hhinc==3) + gm_hhinc3_c*(hhinc==5) + 
    gm_ncar0_c*(ncar==0) + gm_ncar1_c*(ncar==1) + gm_ncar2_c*(ncar==2) + gm_nbic0_c*(nbic==0) + gm_nbic1_c*(nbic==1) + gm_nbic2_c*(nbic==2) + gm_hhC_c*hhC + lm_hab_c*hab
  
  ## Class allocation settings
  classAlloc_settings = list(
    classes = c(class_a = 1, class_b = 2, class_c = 3),
    utilities = V
  )
  
  lcpars[["pi_values"]] = apollo_classAlloc(classAlloc_settings)
  
  return(lcpars)
}

## Group and Validate inputs

apollo_inputs = apollo_validateInputs()

## Define model and likelihood function 

apollo_probabilities = function(apollo_beta, apollo_inputs, functionality = "estimate") {
  
  ## Initialize
  apollo_attach(apollo_beta, apollo_inputs)
  on.exit(apollo_detach(apollo_beta, apollo_inputs))
  
  # Creating list of probabilities P
  
  P = list()
  
  ## Likelihood of indicators to LVs
  
  ol_settings1 = list(outcomeOrdered = att1, 
                      V = zeta_innv1*innv,
                      tau = list(tau_innv1_1, tau_innv1_2, tau_innv1_3, tau_innv1_4),
                      rows = (task==1),
                      componentName = "indic_innv1")
  ol_settings2 = list(outcomeOrdered = att2, 
                      V = zeta_innv2*innv,
                      tau = list(tau_innv2_1, tau_innv2_2, tau_innv2_3, tau_innv2_4),
                      rows = (task==1),
                      componentName = "indic_innv2")
  ol_settings3 = list(outcomeOrdered = att3, 
                      V = zeta_innv3*innv,
                      tau = list(tau_innv3_1, tau_innv3_2, tau_innv3_3, tau_innv3_4),
                      rows = (task==1),
                      componentName = "indic_innv3")
  ol_settings4 = list(outcomeOrdered = att4, 
                      V = zeta_skpt1*skpt,
                      tau = list(tau_skpt1_1, tau_skpt1_2, tau_skpt1_3, tau_skpt1_4),
                      rows = (task==1),
                      componentName = "indic_skpt1")
  ol_settings5 = list(outcomeOrdered = att5, 
                      V = zeta_skpt2*skpt,
                      tau = list(tau_skpt2_1, tau_skpt2_2, tau_skpt2_3, tau_skpt2_4),
                      rows = (task==1),
                      componentName = "indic_skpt2")
  ol_settings6 = list(outcomeOrdered = att10, 
                      V = zeta_skpt3*skpt,
                      tau = list(tau_skpt3_1, tau_skpt3_2, tau_skpt3_3, tau_skpt3_4),
                      rows = (task==1),
                      componentName = "indic_skpt3")
  ol_settings7 = list(outcomeOrdered = att11, 
                      V = zeta_risk1*risk,
                      tau = list(tau_risk1_1, tau_risk1_2, tau_risk1_3, tau_risk1_4),
                      rows = (task==1),
                      componentName = "indic_risk1")
  ol_settings8 = list(outcomeOrdered = att13, 
                      V = zeta_risk2*risk,
                      tau = list(tau_risk2_1, tau_risk2_2, tau_risk2_3, tau_risk2_4),
                      rows = (task==1),
                      componentName = "indic_risk2")
  ol_settings9 = list(outcomeOrdered = att14, 
                      V = zeta_risk3*risk,
                      tau = list(tau_risk3_1, tau_risk3_2, tau_risk3_3, tau_risk3_4),
                      rows = (task==1),
                      componentName = "indic_risk3")
  ol_settings10 = list(outcomeOrdered = att8, 
                      V = zeta_use1*use,
                      tau = list(tau_use1_1, tau_use1_2, tau_use1_3, tau_use1_4),
                      rows = (task==1),
                      componentName = "indic_use1")
  ol_settings11 = list(outcomeOrdered = att16, 
                       V = zeta_use2*use,
                       tau = list(tau_use2_1, tau_use2_2, tau_use2_3, tau_use2_4),
                       rows = (task==1),
                       componentName = "indic_use2")
  ol_settings12 = list(outcomeOrdered = att17, 
                       V = zeta_use3*use,
                       tau = list(tau_use3_1, tau_use3_2, tau_use3_3, tau_use3_4),
                       rows = (task==1),
                       componentName = "indic_use3")
  ol_settings13 = list(outcomeOrdered = att18, 
                       V = zeta_use4*use,
                       tau = list(tau_use4_1, tau_use4_2, tau_use4_3, tau_use4_4),
                       rows = (task==1),
                       componentName = "indic_use4")
  ol_settings14 = list(outcomeOrdered = att19, 
                       V = zeta_use5*use,
                       tau = list(tau_use5_1, tau_use5_2, tau_use5_3, tau_use5_4),
                       rows = (task==1),
                       componentName = "indic_use5")
  ol_settings15 = list(outcomeOrdered = att21, 
                       V = zeta_sn1*sn,
                       tau = list(tau_sn1_1, tau_sn1_2, tau_sn1_3, tau_sn1_4),
                       rows = (task==1),
                       componentName = "indic_sn1")
  ol_settings16 = list(outcomeOrdered = att22, 
                       V = zeta_sn2*sn,
                       tau = list(tau_sn2_1, tau_sn2_2, tau_sn2_3, tau_sn2_4),
                       rows = (task==1),
                       componentName = "indic_sn2")
  ol_settings17 = list(outcomeOrdered = att23, 
                       V = zeta_env1*env,
                       tau = list(tau_env1_1, tau_env1_2, tau_env1_3, tau_env1_4),
                       rows = (task==1),
                       componentName = "indic_env1")
  ol_settings18 = list(outcomeOrdered = att24, 
                       V = zeta_env2*env,
                       tau = list(tau_env2_1, tau_env2_2, tau_env2_3, tau_env2_4),
                       rows = (task==1),
                       componentName = "indic_env2")
  ol_settings19 = list(outcomeOrdered = att25, 
                       V = zeta_env3*env,
                       tau = list(tau_env3_1, tau_env3_2, tau_env3_3, tau_env3_4),
                       rows = (task==1),
                       componentName = "indic_env3")
  ol_settings20 = list(outcomeOrdered = hab1, 
                       V = zeta_hab1*hab,
                       tau = list(tau_hab1_1, tau_hab1_2, tau_hab1_3, tau_hab1_4),
                       rows = (task==1),
                       componentName = "indic_hab1")
  ol_settings21 = list(outcomeOrdered = hab2, 
                       V = zeta_hab2*hab,
                       tau = list(tau_hab2_1, tau_hab2_2, tau_hab2_3, tau_hab2_4),
                       rows = (task==1),
                       componentName = "indic_hab2")
  ol_settings22 = list(outcomeOrdered = hab3, 
                       V = zeta_hab3*hab,
                       tau = list(tau_hab3_1, tau_hab3_2, tau_hab3_3, tau_hab3_4),
                       rows = (task==1),
                       componentName = "indic_hab3")
  ol_settings23 = list(outcomeOrdered = hab4, 
                       V = zeta_hab4*hab,
                       tau = list(tau_hab4_1, tau_hab4_2, tau_hab4_3, tau_hab4_4),
                       rows = (task==1),
                       componentName = "indic_hab4")
  ol_settings24 = list(outcomeOrdered = hab5, 
                       V = zeta_hab5*hab,
                       tau = list(tau_hab5_1, tau_hab5_2, tau_hab5_3, tau_hab5_4),
                       rows = (task==1),
                       componentName = "indic_hab5")
  
  P[["indic_innv1"]] = apollo_ol(ol_settings1, functionality)
  P[["indic_innv2"]] = apollo_ol(ol_settings2, functionality)
  P[["indic_innv3"]] = apollo_ol(ol_settings3, functionality)
  P[["indic_skpt1"]] = apollo_ol(ol_settings4, functionality)
  P[["indic_skpt2"]] = apollo_ol(ol_settings5, functionality)
  P[["indic_skpt3"]] = apollo_ol(ol_settings6, functionality)
  P[["indic_risk1"]] = apollo_ol(ol_settings7, functionality)
  P[["indic_risk2"]] = apollo_ol(ol_settings8, functionality)
  P[["indic_risk3"]] = apollo_ol(ol_settings9, functionality)
  P[["indic_use1"]] = apollo_ol(ol_settings10, functionality)
  P[["indic_use2"]] = apollo_ol(ol_settings11, functionality)
  P[["indic_use3"]] = apollo_ol(ol_settings12, functionality)
  P[["indic_use4"]] = apollo_ol(ol_settings13, functionality)
  P[["indic_use5"]] = apollo_ol(ol_settings14, functionality)
  P[["indic_sn1"]] = apollo_ol(ol_settings15, functionality)
  P[["indic_sn2"]] = apollo_ol(ol_settings16, functionality)
  P[["indic_env1"]] = apollo_ol(ol_settings17, functionality)
  P[["indic_env2"]] = apollo_ol(ol_settings18, functionality)
  P[["indic_env3"]] = apollo_ol(ol_settings19, functionality)
  P[["indic_hab1"]] = apollo_ol(ol_settings20, functionality)
  P[["indic_hab2"]] = apollo_ol(ol_settings21, functionality)
  P[["indic_hab3"]] = apollo_ol(ol_settings22, functionality)
  P[["indic_hab4"]] = apollo_ol(ol_settings23, functionality)
  P[["indic_hab5"]] = apollo_ol(ol_settings24, functionality)
  
  P[["indic_innv1"]] = apollo_panelProd(P[["indic_innv1"]], apollo_inputs, functionality)
  P[["indic_innv2"]] = apollo_panelProd(P[["indic_innv2"]], apollo_inputs, functionality)
  P[["indic_innv3"]] = apollo_panelProd(P[["indic_innv3"]], apollo_inputs, functionality)
  P[["indic_skpt1"]] = apollo_panelProd(P[["indic_skpt1"]], apollo_inputs, functionality)
  P[["indic_skpt2"]] = apollo_panelProd(P[["indic_skpt2"]], apollo_inputs, functionality)
  P[["indic_skpt3"]] = apollo_panelProd(P[["indic_skpt3"]], apollo_inputs, functionality)
  P[["indic_risk1"]] = apollo_panelProd(P[["indic_risk1"]], apollo_inputs, functionality)
  P[["indic_risk2"]] = apollo_panelProd(P[["indic_risk2"]], apollo_inputs, functionality) 
  P[["indic_risk3"]] = apollo_panelProd(P[["indic_risk3"]], apollo_inputs, functionality)
  P[["indic_use1"]] = apollo_panelProd(P[["indic_use1"]], apollo_inputs, functionality)
  P[["indic_use2"]] = apollo_panelProd(P[["indic_use2"]], apollo_inputs, functionality)
  P[["indic_use3"]] = apollo_panelProd(P[["indic_use3"]], apollo_inputs, functionality)
  P[["indic_use4"]] = apollo_panelProd(P[["indic_use4"]], apollo_inputs, functionality)
  P[["indic_use5"]] = apollo_panelProd(P[["indic_use5"]], apollo_inputs, functionality)
  P[["indic_sn1"]] = apollo_panelProd(P[["indic_sn1"]], apollo_inputs, functionality)
  P[["indic_sn2"]] = apollo_panelProd(P[["indic_sn2"]], apollo_inputs, functionality)
  P[["indic_env1"]] = apollo_panelProd(P[["indic_env1"]], apollo_inputs, functionality)
  P[["indic_env2"]] = apollo_panelProd(P[["indic_env2"]], apollo_inputs, functionality)
  P[["indic_env3"]] = apollo_panelProd(P[["indic_env3"]], apollo_inputs, functionality)
  P[["indic_hab1"]] = apollo_panelProd(P[["indic_hab1"]], apollo_inputs, functionality)
  P[["indic_hab2"]] = apollo_panelProd(P[["indic_hab2"]], apollo_inputs, functionality)
  P[["indic_hab3"]] = apollo_panelProd(P[["indic_hab3"]], apollo_inputs, functionality)
  P[["indic_hab4"]] = apollo_panelProd(P[["indic_hab4"]], apollo_inputs, functionality)
  P[["indic_hab5"]] = apollo_panelProd(P[["indic_hab5"]], apollo_inputs, functionality)
  
  ## Compute class-specific utilities in choice sub-model
  ## Loop over classes
  
  for (s in 1:3) {
    V = list()
    
    V[["car"]] = asc_car[[s]] + b_car_ivtt[[s]]*car_ivtt + b_car_egt[[s]]*car_egt + b_car_tc[[s]]*car_tc + b_car_pc[[s]]*car_pc + 
      lm_innv_car[[s]]*innv + lm_skpt_car[[s]]*skpt + lm_risk_car[[s]]*risk + lm_use_car[[s]]*use + lm_sn_car[[s]]*sn + lm_env_car[[s]]*env 
    V[["pt"]] = asc_pt[[s]] + b_pt_ivtt[[s]]*pt_ivtt + b_pt_act[[s]]*pt_act + b_pt_egt[[s]]*pt_egt + b_pt_wt[[s]]*pt_wt + b_pt_tc[[s]]*pt_tc +
      lm_innv_pt[[s]]*innv + lm_skpt_pt[[s]]*skpt + lm_risk_pt[[s]]*risk + lm_use_pt[[s]]*use + lm_sn_pt[[s]]*sn + lm_env_pt[[s]]*env 
    V[["ecar"]] = asc_ecar[[s]] + b_ecar_ivtt[[s]]*ecar_ivtt + b_ecar_act[[s]]*ecar_act + b_ecar_egt[[s]]*ecar_egt + b_ecar_wt[[s]]*ecar_wt + b_ecar_tc[[s]]*ecar_tc + b_ecar_av[[s]]*ecar_av +
      lm_innv_ecar[[s]]*innv + lm_skpt_ecar[[s]]*skpt + lm_risk_ecar[[s]]*risk + lm_use_ecar[[s]]*use + lm_sn_ecar[[s]]*sn + lm_env_ecar[[s]]*env 
    V[["ebike"]] = asc_ebike[[s]] + b_ebike_ivtt[[s]]*ebike_ivtt + b_ebike_act[[s]]*ebike_act + b_ebike_egt[[s]]*ebike_egt + b_ebike_wt[[s]]*ebike_wt + b_ebike_tc[[s]]*ebike_tc + b_ebike_av[[s]]*ebike_av +
      lm_innv_ebike[[s]]*innv + lm_skpt_ebike[[s]]*skpt + lm_risk_ebike[[s]]*risk + lm_use_ebike[[s]]*use + lm_sn_ebike[[s]]*sn + lm_env_ebike[[s]]*env 
    V[["escoot"]] = asc_escoot[[s]] + b_escoot_ivtt[[s]]*escoot_ivtt + b_escoot_act[[s]]*escoot_act + b_escoot_egt[[s]]*escoot_egt + b_escoot_wt[[s]]*escoot_wt + b_escoot_tc[[s]]*escoot_tc + b_escoot_av[[s]]*escoot_av +
      lm_innv_escoot[[s]]*innv + lm_skpt_escoot[[s]]*skpt + lm_risk_escoot[[s]]*risk + lm_use_escoot[[s]]*use + lm_sn_escoot[[s]]*sn + lm_env_escoot[[s]]*env 
    V[["aebike"]] = asc_aebike[[s]] + b_aebike_ivtt[[s]]*aebike_ivtt + b_aebike_wk[[s]]*aebike_wk + b_aebike_wt[[s]]*aebike_wt + b_aebike_tc[[s]]*aebike_tc + b_aebike_av[[s]]*aebike_av +
      lm_innv_aebike[[s]]*innv + lm_skpt_aebike[[s]]*skpt + lm_risk_aebike[[s]]*risk + lm_use_aebike[[s]]*use + lm_sn_aebike[[s]]*sn + lm_env_aebike[[s]]*env 
    V[["aescoot"]] = asc_aescoot[[s]] + b_aescoot_ivtt[[s]]*aescoot_ivtt + b_aescoot_wk[[s]]*aescoot_wk + b_aescoot_wt[[s]]*aescoot_wt + b_aescoot_tc[[s]]*aescoot_tc + b_aescoot_av[[s]]*aescoot_av +
      lm_innv_aescoot[[s]]*innv + lm_skpt_aescoot[[s]]*skpt + lm_risk_aescoot[[s]]*risk + lm_use_aescoot[[s]]*use + lm_sn_aescoot[[s]]*sn + lm_env_aescoot[[s]]*env 
    V[["awalk"]] = asc_awalk[[s]] + b_awalk[[s]]*awalk + lm_risk_awalk[[s]]*risk + lm_use_awalk[[s]]*use
    
    ## Settings for MNL model component
    mnl_settings = list(
      alternatives = c(car=1, pt=2, ecar=3, ebike=4, escoot=5, aebike=6, aescoot=7, awalk=8), 
      avail = list(car=av_car, pt=av_pt, ecar=av_ecar, ebike=av_ebike, escoot=av_escoot, aebike=av_aebike, aescoot=av_aescoot, awalk=av_awalk), 
      choiceVar = mode,
      utilities = V,
      componentName = paste0("class_", s)
    )
    
    ## Compute within-class choice probabilities using MNL
    P[[paste0("class_",s)]] = apollo_mnl(mnl_settings, functionality)
    
    ## Product across observations for same ID
    P[[paste0("class_",s)]] = apollo_panelProd(P[[paste0("class_",s)]], apollo_inputs, functionality)
  }
  
  ### Compute latent class model probabilities
  lc_settings  = list(inClassProb = P[paste0("class_", 1:3)], classProb=pi_values, componentName="choice")
  P[["choice"]] = apollo_lc(lc_settings, apollo_inputs, functionality)
  
  ### Comment out as necessary
  P = apollo_combineModels(P, apollo_inputs, functionality, components=c("indic_innv1", "indic_innv2", "indic_innv3", 
                                                                         "indic_skpt1", "indic_skpt2", "indic_skpt3",
                                                                         "indic_risk1", "indic_risk2", "indic_risk3",
                                                                         "indic_use1", "indic_use2", "indic_use3", "indic_use4", "indic_use5",
                                                                         "indic_sn1", "indic_sn2",
                                                                         "indic_env1", "indic_env2", "indic_env3",
                                                                         "indic_hab1", "indic_hab2", "indic_hab3", "indic_hab4", "indic_hab5", 
                                                                         "choice"))
  ## Average across inter-individual draws
  P = apollo_avgInterDraws(P, apollo_inputs, functionality)
  ## Prepare and return outputs of function
  P = apollo_prepareProb(P, apollo_inputs, functionality)
  return(P)
} 

## Calculate LL at starting values

apollo_llCalc(apollo_beta, apollo_probabilities, apollo_inputs)

## Model Estimation

model = apollo_estimate(apollo_beta, apollo_fixed, apollo_probabilities, apollo_inputs, estimate_settings = list(maxIterations=100))

## Model output

apollo_modelOutput(model)
stephanehess
Site Admin
Posts: 1235
Joined: 24 Apr 2020, 16:29

Re: speedTest function in multiple cores

Post by stephanehess »

Hi

your model is complex and you have a very large number of parameters. Other than using a more powerful computer, one option to consider would be continuous as opposed to ordinal measurement models. This would mean fewer parameters, but is of course a simplification

Stephane
--------------------------------
Stephane Hess
www.stephanehess.me.uk
Post Reply