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. This may take a day or two at busy times. There is no need to submit the post multiple times.

LCCM model with two choice models

Ask questions about model specifications. Ideally include a mathematical explanation of your proposed model.
Post Reply
jaimlk77
Posts: 2
Joined: 04 Mar 2021, 04:15

LCCM model with two choice models

Post by jaimlk77 »

Hi

I am trying to estimate a LCCM with two choice models in each latent class. Is that possible to estimate in Apollo?

As shown in the attached subset of the R code. I am trying to wrap two MNLs in latent classes. However, I keep getting the following error:

Error in apollo_lc(lc_settings, apollo_inputs, functionality) :
Arguments 'inClassProb' and 'classProb' for model component "LC" must have the same length.


I thought using the apollo_combineModels would resolve this issue but that did not help. Does anyone have any suggestions on how I can resolve this issue?

Thanks,
Jai




### Loop over classes
for(s in 1:2){

### Compute class-specific utilities
V_1[["alt_1"]] = asc_1_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_2"]] = asc_2_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_3"]] = asc_3_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_4"]] = asc_4_1[[s]]

mnl_settings_1$utilities = V_1
mnl_settings_1$componentName = paste0("model1_Class_",s)

### Compute within-class choice probabilities using MNL model
P[[paste0("model1_Class_",s)]] = apollo_mnl(mnl_settings_1, functionality)

V_2 = list()

V_2[["alt_1"]] = asc_1_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_2"]] = asc_2_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_3"]] = asc_3_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_4"]] = asc_4_2[[s]]

mnl_settings_2$utilities = V_2
mnl_settings_2$componentName = paste0("model2_Class_",s)

### Compute within-class choice probabilities using MNL model
P[[paste0("model2_Class_",s)]] = apollo_mnl(mnl_settings_2, functionality)

}
stephanehess
Site Admin
Posts: 974
Joined: 24 Apr 2020, 16:29

Re: LCCM model with two choice models

Post by stephanehess »

Hi

could you please post the entire apollo_probabilities function for your example?

Thanks

Stephane
--------------------------------
Stephane Hess
www.stephanehess.me.uk
jaimlk77
Posts: 2
Joined: 04 Mar 2021, 04:15

Re: LCCM model with two choice models

Post by jaimlk77 »

Hi

Please see the entire apollo_probabilities below.

Thanks,
Jai

apollo_probabilities=function(apollo_beta, apollo_inputs, functionality="estimate"){

### Attach inputs and detach after function exit
apollo_attach(apollo_beta, apollo_inputs)
on.exit(apollo_detach(apollo_beta, apollo_inputs))

### Create list of probabilities P
P = list()

### List of utilities: these must use the same names as in mnl_settings, order is irrelevant
V_1 = list()

### Define settings for MNL model component
mnl_settings_1 = list(
alternatives = c(alt_1 = 1, alt_2 = 2, alt_3 = 3, alt_4 = 4),
avail = list(alt_1 = 1, alt_2 = 1, alt_3 = 1, alt_4 = 1),
choiceVar = actual_veh_change_fall19_fall20)

### Define settings for MNL model component
mnl_settings_2 = list(
alternatives = c(alt_1 = 1, alt_2 = 2, alt_3 = 3, alt_4 = 4),
avail = list(alt_1 = 1, alt_2 = 1, alt_3 = 1, alt_4 = 1),
choiceVar = expect_change_fall20_summer20)
### Loop over classes
for(s in 1:2){

### Compute class-specific utilities
V_1[["alt_1"]] = asc_1_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_2"]] = asc_2_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_3"]] = asc_3_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_4"]] = asc_4_1[[s]]

mnl_settings_1$utilities = V_1
mnl_settings_1$componentName = paste0("model1_Class_",s)

### Compute within-class choice probabilities using MNL model
P[[paste0("model1_Class_",s)]] = apollo_mnl(mnl_settings_1, functionality)

V_2 = list()

V_2[["alt_1"]] = asc_1_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_2"]] = asc_2_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_3"]] = asc_3_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_4"]] = asc_4_2[[s]]

mnl_settings_2$utilities = V_2
mnl_settings_2$componentName = paste0("model2_Class_",s)

### Compute within-class choice probabilities using MNL model
P[[paste0("model2_Class_",s)]] = apollo_mnl(mnl_settings_2, functionality)

}

### Compute latent class model probabilities
lc_settings = list(inClassProb = P, classProb=pi_values)


P[["lc_model"]] = apollo_lc(lc_settings, apollo_inputs, functionality)
### Take product across observation for same individual
# P = apollo_panelProd(P, apollo_inputs, functionality)

#P = apollo_combineModels(P, apollo_inputs, functionality)

### Prepare and return outputs of function
P = apollo_prepareProb(P, apollo_inputs, functionality)
return(P)
}
stephanehess
Site Admin
Posts: 974
Joined: 24 Apr 2020, 16:29

Re: LCCM model with two choice models

Post by stephanehess »

Hi

you'll need to call apollo_combineModels inside the loop over classes, so that your two MNL models are combined inside the class.

But I think you also need to change some names for that to work. Can you try the following:

apollo_probabilities=function(apollo_beta, apollo_inputs, functionality="estimate"){

### Attach inputs and detach after function exit
apollo_attach(apollo_beta, apollo_inputs)
on.exit(apollo_detach(apollo_beta, apollo_inputs))

### Create list of probabilities P
P = list()

### List of utilities: these must use the same names as in mnl_settings, order is irrelevant
V_1 = list()

### Define settings for MNL model component
mnl_settings_1 = list(
alternatives = c(alt_1 = 1, alt_2 = 2, alt_3 = 3, alt_4 = 4),
avail = list(alt_1 = 1, alt_2 = 1, alt_3 = 1, alt_4 = 1),
choiceVar = actual_veh_change_fall19_fall20)

### Define settings for MNL model component
mnl_settings_2 = list(
alternatives = c(alt_1 = 1, alt_2 = 2, alt_3 = 3, alt_4 = 4),
avail = list(alt_1 = 1, alt_2 = 1, alt_3 = 1, alt_4 = 1),
choiceVar = expect_change_fall20_summer20)
### Loop over classes
for(s in 1:2){

P_temp=list()

### Compute class-specific utilities
V_1[["alt_1"]] = asc_1_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_2"]] = asc_2_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_3"]] = asc_3_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_4"]] = asc_4_1[[s]]

mnl_settings_1$utilities = V_1
mnl_settings_1$componentName = paste0("model1_Class_",s)

### Compute within-class choice probabilities using MNL model
P_temp[[paste0("model1_Class_",s)]] = apollo_mnl(mnl_settings_1, functionality)

V_2 = list()

V_2[["alt_1"]] = asc_1_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_2"]] = asc_2_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_3"]] = asc_3_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_4"]] = asc_4_2[[s]]

mnl_settings_2$utilities = V_2
mnl_settings_2$componentName = paste0("model2_Class_",s)

### Compute within-class choice probabilities using MNL model
P_temp[[paste0("model2_Class_",s)]] = apollo_mnl(mnl_settings_2, functionality)

P[[paste0("Class_",s)]] = apollo_combineModels(P_temp,functionality)

}

### Compute latent class model probabilities
lc_settings = list(inClassProb = P, classProb=pi_values)


P[["lc_model"]] = apollo_lc(lc_settings, apollo_inputs, functionality)
### Take product across observation for same individual
P = apollo_panelProd(P, apollo_inputs, functionality)

#P = apollo_combineModels(P, apollo_inputs, functionality)

### Prepare and return outputs of function
P = apollo_prepareProb(P, apollo_inputs, functionality)
return(P)
}
Re: LCCM model with two choice models
by stephanehess » Mon Mar 21, 2022 7:28 pm

Hi

could you please post the entire apollo_probabilities function for your example?

Thanks

Stephane
LCCM model with two choice models
by jaimlk77 » Fri Mar 18, 2022 6:12 pm

Hi

I am trying to estimate a LCCM with two choice models in each latent class. Is that possible to estimate in Apollo?

As shown in the attached subset of the R code. I am trying to wrap two MNLs in latent classes. However, I keep getting the following error:

Error in apollo_lc(lc_settings, apollo_inputs, functionality) :
Arguments 'inClassProb' and 'classProb' for model component "LC" must have the same length.

I thought using the apollo_combineModels would resolve this issue but that did not help. Does anyone have any suggestions on how I can resolve this issue?

Thanks,
Jai




### Loop over classes
for(s in 1:2){

### Compute class-specific utilities
V_1[["alt_1"]] = asc_1_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_2"]] = asc_2_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_3"]] = asc_3_1[[s]] + b_age_num_1_1[[s]] * (age_num == 1) + b_age_num_2_1[[s]] * (age_num == 2) + b_gender_1[[s]] * (gender_num == 1)
V_1[["alt_4"]] = asc_4_1[[s]]

mnl_settings_1$utilities = V_1
mnl_settings_1$componentName = paste0("model1_Class_",s)

### Compute within-class choice probabilities using MNL model
P[[paste0("model1_Class_",s)]] = apollo_mnl(mnl_settings_1, functionality)

V_2 = list()

V_2[["alt_1"]] = asc_1_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_2"]] = asc_2_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_3"]] = asc_3_2[[s]] + b_age_num_1_2[[s]] * (age_num == 1) + b_age_num_2_2[[s]] * (age_num == 2) + b_gender_2[[s]] * (gender_num == 1)
V_2[["alt_4"]] = asc_4_2[[s]]

mnl_settings_2$utilities = V_2
mnl_settings_2$componentName = paste0("model2_Class_",s)

### Compute within-class choice probabilities using MNL model
P[[paste0("model2_Class_",s)]] = apollo_mnl(mnl_settings_2, functionality)

}
--------------------------------
Stephane Hess
www.stephanehess.me.uk
punyabeet
Posts: 19
Joined: 08 May 2020, 11:43

Re: LCCM model with two choice models

Post by punyabeet »

Hi Prof Stephane

Thanks for the apollo forum. I tried to estimate LCCM with two choice models in each latent class with a similar specification as above. However, I keep getting the following error:

Error in apollo_validate(classAlloc_settings, modelType, functionality, :
Some of the names of the classes for model component 'classAlloc' do not match!

I think I have made an error in defining the latent class components. Can you please point out the mistake in the R code snippet?

# ################################################################# #
#### DEFINE MODEL PARAMETERS ####
# ################################################################# #

### Vector of parameters, including any that are kept fixed in estimation
apollo_beta = c(asc_1_1 = 0,
asc_2_1 = 0,
asc_3_1 = 0,
asc_1_2 = 0,
asc_2_2 = 0,
asc_3_2 = 0,
b_tt_a = 0,
b_tt_b = 0,
delta_a_1 = 0,
delta_a_2 = 0,
delta_b_1 = 0,
delta_b_2 = 0)

### Vector with names (in quotes) of parameters to be kept fixed at their starting value in apollo_beta, use apollo_beta_fixed = c() if none
apollo_fixed = c("delta_b_1", "delta_b_2", "asc_3_1", "asc_3_2")

# ################################################################# #
#### DEFINE LATENT CLASS COMPONENTS ####
# ################################################################# #

apollo_lcPars=function(apollo_beta, apollo_inputs){
lcpars = list()
lcpars[["b_tt_1"]] = list(b_tt_a, b_tt_b)
lcpars[["b_tt_2"]] = list(b_tt_a, b_tt_b)

V_1 = list()
V_1[["class_a"]] = delta_a_1
V_1[["class_b"]] = delta_b_1

V_2 = list()
V_2[["class_a"]] = delta_a_2
V_2[["class_b"]] = delta_b_2


### Settings for class allocation models
classAlloc_settings = list(
classes = c(class_a=1, class_b=2),
utilities = V
)

### Settings for class allocation models
classAlloc_settings = list(
classes = c(class_a = 1, class_b = 2),
utilities = c(V_1, V_2),
avail = 1
)

lcpars[["pi_values"]] = apollo_classAlloc(classAlloc_settings)

return(lcpars)
}

Many thanks for considering my request.
Punya
dpalma
Posts: 190
Joined: 24 Apr 2020, 17:54

Re: LCCM model with two choice models

Post by dpalma »

Hi Punya,

Your code is using two class allocation functions. What I believe you want is a single allocation function, and within each class to have two choices. Below is an example that does so.

The example uses the example database apollo_timeUseData, and it estimates a latent class model with two classes. Within each class, there are two choices: to engage in leasure (yes/no), and to engage in work (yes/no). Each choice has different parameters inside each class. You can try using this structure for your application.

Cheers
David

Code: Select all

# ################################################################# #
#### LOAD LIBRARY AND DEFINE CORE SETTINGS                       ####
# ################################################################# #

### Initialise
rm(list = ls())
library(apollo)
apollo_initialise()

### Set core controls
apollo_control = list(
  modelName       = "LC_twoChoices",
  modelDescr      = "LC models with 2 classes and two choices inside each",
  indivID         = "indivID"
)

# ################################################################# #
#### LOAD DATA AND APPLY ANY TRANSFORMATIONS                     ####
# ################################################################# #

### Loading data from package
### if data is to be loaded from a file (e.g. called data.csv), 
### the code would be: database = read.csv("data.csv",header=TRUE)
database = apollo_timeUseData
### for data dictionary, use ?apollo_timeUseData
# outside good: time spent at home and travelling
database$t_outs = rowSums(database[,c("t_a01", "t_a06", "t_a10", "t_a11", "t_a12")])/60 
database$t_leis = rowSums(database[,c("t_a07", "t_a08", "t_a09")])/60
database$t_work = database$t_a02/60
database$t_scho = database$t_a03/60
database$t_shop = database$t_a04/60
database$t_priv = database$t_a05/60
database$doesLeisure <- database$t_leis > 0
database$doesWork    <- database$t_work > 0

# ################################################################# #
#### DEFINE MODEL PARAMETERS                                     ####
# ################################################################# #

### Vector of parameters, including any that are kept fixed in estimation
apollo_beta = c(ascL_1 = 0, ascL_2 = 0, 
                bFeL_1 = 0, bFeL_2 = 0, 
                bAgL_1 = 0, bAgL_2 = 0, 
                bFTL_1 = 0, bFTL_2 = 0, 
                bWkL_1 = 0, bWkL_2 = 0, 
                ascW_1 = 0, ascW_2 = 0, 
                bFeW_1 = 0, bFeW_2 = 0, 
                bAgW_1 = 0, bAgW_2 = 0, 
                bFTW_1 = 0, bFTW_2 = 0, 
                bWkW_1 = 0, bWkW_2 = 0, 
                delt_1 = 0, delt_2 = 0)

### Vector with names (in quotes) of parameters to be kept fixed at their
### starting value in apollo_beta, use apollo_beta_fixed = c() if none.
apollo_fixed = c("delt_2")

# ################################################################# #
#### DEFINE LATENT CLASS COMPONENTS                              ####
# ################################################################# #

apollo_lcPars=function(apollo_beta, apollo_inputs){
  lcpars = list()
  lcpars[["ascL"]] = list(ascL_1, ascL_2)
  lcpars[["bFeL"]] = list(bFeL_1, bFeL_2)
  lcpars[["bAgL"]] = list(bAgL_1, bAgL_2)
  lcpars[["bFTL"]] = list(bFTL_1, bFTL_2)
  lcpars[["bWkL"]] = list(bWkL_1, bWkL_2)
  lcpars[["ascW"]] = list(ascW_1, ascW_2)
  lcpars[["bFeW"]] = list(bFeW_1, bFeW_2)
  lcpars[["bAgW"]] = list(bAgW_1, bAgW_2)
  lcpars[["bFTW"]] = list(bFTW_1, bFTW_2)
  lcpars[["bWkW"]] = list(bWkW_1, bWkW_2)
  
  classAlloc_settings = list(
    utilities = list(class_1 = delt_1, 
                     class_2 = delt_2)
  )
  
  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"){
  
  ### Initialise
  apollo_attach(apollo_beta, apollo_inputs)
  on.exit(apollo_detach(apollo_beta, apollo_inputs))
  P = list()
  
  ### Define settings for MNL model component that are generic across classes
  mnl_leisure_settings = list(
    alternatives = c(no=0, yes=1),
    choiceVar    = doesLeisure
  )
  mnl_work_settings = list(
    alternatives = c(no=0, yes=1),
    choiceVar    = doesWork
  )
  
  ### Loop over classes
  for(s in 1:2){
    
    ### Leisure decision within class s
    mnl_leisure_settings$utilities = list(
      no  = 0, 
      yes = ascL[[s]] + bFeL[[s]]*female + bAgL[[s]]*age + bFTL[[s]]*occ_full_time + bWkL[[s]]*weekend
    )
    P[[paste0("Class_",s,"_leisure")]] = apollo_mnl(mnl_leisure_settings, functionality)
    
    ### Work decision within class s
    mnl_work_settings$utilities = list(
      no  = 0, 
      yes = ascW[[s]] + bFeW[[s]]*female + bAgW[[s]]*age + bFTW[[s]]*occ_full_time + bWkW[[s]]*weekend
    )
    P[[paste0("Class_",s,"_work")]] = apollo_mnl(mnl_work_settings, functionality)
    
    ### Take product of both decisions
    P[[paste0("Class_",s)]] = P[[paste0("Class_",s,"_leisure")]]*P[[paste0("Class_",s,"_work")]]
    
    ### Take product across observation for same individual
    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_",1:2)], classProb=pi_values)
  P[["model"]] = apollo_lc(lc_settings, apollo_inputs, functionality)
  
  ### Prepare and return outputs of function
  P = apollo_prepareProb(P, apollo_inputs, functionality)
  return(P)
}

# ################################################################# #
#### MODEL ESTIMATION AND OUTPUT                                 ####
# ################################################################# #

### Estimation
model = apollo_estimate(apollo_beta, apollo_fixed, apollo_probabilities, 
                        apollo_inputs)

### Output to screen
apollo_modelOutput(model)

### Output to file
apollo_saveOutput(model)
Post Reply