Hello!
I have a problem with the estimation of latent class model with random parameters.
When I use this model, I get an (as I think) incorrect Rho- square and Adj.Rho- square values.
I tried to run the example 21. from the website but I get the same result, which is the following:
"LL(start) : -1698.575
LL(0, whole model) : -242047
LL(final, whole model) : -1502.223
Rho-square (0) : 0.9938
Adj.Rho-square (0) : 0.9937
AIC : 3034.45
BIC : 3126.82
LL(0,Class_1) : -2420.47
LL(final,Class_1) : -2675.017
LL(0,Class_2) : -2420.47
LL(final,Class_2) : -1637.6
Estimated parameters : 15".
Can the problem come from the irreal value of LL(0)?
Best regards,
Peter
Important: Read this before posting to this forum
- 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.
- 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
- Before asking a question on the forum, users are kindly requested to follow these steps:
- Check that the same issue has not already been addressed in the forum - there is a search tool.
- 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
- Check the frequently asked questions section on the Apollo website, which discusses some common issues/failures. Please see http://www.apollochoicemodelling.com/faq.html
- Make sure that R is using the latest official release of Apollo.
- Users can check which version they are running by entering packageVersion("apollo").
- Then check what is the latest full release (not development version) at http://www.ApolloChoiceModelling.com/code.html.
- To update to the latest official version, just enter install.packages("apollo"). To update to a development version, download the appropriate binary file from http://www.ApolloChoiceModelling.com/code.html, and install the package from file
- If the above steps do not resolve the issue, then users should follow these steps when posting a question:
- provide full details on the issue, including the entire code and output, including any error messages
- 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.
Latent class with continuous random parameters: LL0 and rho2
-
- Site Admin
- Posts: 1050
- Joined: 24 Apr 2020, 16:29
Re: Latent class with continuous random parameters
Peter
you found a bug - thanks! This was very specifically happening in the case of latent class models with continuous random coefficients in the class allocation component. I've just uploaded a new version of Apollo 0.1.1 on the website (www.ApolloChoiceModelling.com/code.html)
Stephane
you found a bug - thanks! This was very specifically happening in the case of latent class models with continuous random coefficients in the class allocation component. I've just uploaded a new version of Apollo 0.1.1 on the website (www.ApolloChoiceModelling.com/code.html)
Stephane
-
- Posts: 11
- Joined: 24 Aug 2020, 15:07
- Location: Brazil
- Contact:
Re: Latent class with continuous random parameters: LL0 and rho2
Hi professor Hess,
I'm estimating a Latent Class Model with a latent variable in the class alocation model. I installed version 0.1.1 but I had the same problem of Peter.
Following the code of the model:
The result:
Gabriel
I'm estimating a Latent Class Model with a latent variable in the class alocation model. I installed version 0.1.1 but I had the same problem of Peter.
Following the code of the model:
Code: Select all
#working directory
setwd("C:\\Users\\calde\\Google Drive\\Mestrado 2019_Gabriel Caldeira\\Modelos\\OD\\Banco de dados")
library(apollo)
library(tidyverse)
rm(list = ls())
parallel::detectCores() #cores available
#read database
database <- read.csv(file = paste0(getwd(),'\\cleaning\\database_calib.csv'))
database <- database %>%
mutate(picomanha = case_when(h_saida >= 6 & h_saida <= 9 ~1,TRUE~0))
###Preparing environment
apollo_initialise()
###Options controlling the running of the code
apollo_control = list(
modelName ="mnl_lc18",
modelDescr = "latent class model",
indivID ="id_pess",
mixing = T,
nCores = 10
)
#parameters
apollo_beta <- c(asc1 = 0,
asc2 = 0,
asc3 = 0,
#class 1
b_ttauto_1 = 0,
b_ttbus_1 = 0,
b_ttmetro_1 = 0,
b_coauto_1 = 0,
b_cometro_1 = 0,
#class 2
b_ttauto_2 = 0,
b_ttbus_2 = 0,
b_ttmetro_2 = 0,
b_coauto_2 = 0,
b_cometro_2 = 0,
#class membership
s_1 = 0,
s_2 = 0,
b_sexo_al = 0,
b_idade_al = 0,
b_nomorad_al = 0,
b_totviag_al = 0,
b_grau1_al = 0,
b_grau2_al = 0,
b_pico_al = 0,
param = 0)
#fixed parameters
apollo_fixed <- c('asc1','s_1')
#starting values
setwd('C:\\Users\\calde\\Google Drive\\Mestrado 2019_Gabriel Caldeira\\JOCM\\modelo')
apollo_beta <- apollo_readBeta(apollo_beta,apollo_fixed,'mnl_lc10',overwriteFixed = FALSE)
#criando draws
apollo_draws = list(
interDrawsType = 'halton',
interNDraws = 300,
interNormDraws = c('eta')
)
apollo_randCoeff <- function(apollo_beta,apollo_inputs){
randcoeff = list()
randcoeff[['LV']] = b_sexo_al*sexod + b_idade_al*idade + b_nomorad_al*no_morad + b_totviag_al*tot_viag +
b_grau1_al*grau1 + b_grau2_al*grau2 + b_pico_al*picomanha + eta
return(randcoeff)
}
###Grouping latent class parameters
apollo_lcPars = function(apollo_beta, apollo_inputs){
lcpars = list()
lcpars[['b_ttauto']] = list(b_ttauto_1,b_ttauto_2)
lcpars[['b_ttbus']] = list(b_ttbus_1,b_ttbus_2)
lcpars[['b_ttmetro']] = list(b_ttmetro_1,b_ttmetro_2)
lcpars[['b_coauto']] = list(b_coauto_1,b_coauto_2)
lcpars[['b_cometro']] = list(b_cometro_1,b_cometro_2)
###Class membership probabilities based on s_1, s_2: use of MNL fomrula
V=list()
V[["class_1"]] = s_1
V[["class_2"]] = s_2 + param*LV
###Settings for class membership probabilities
mnl_settings = list(
alternatives = c(class_1=1, class_2=2),
avail = 1,
choiceVar = NA,
V = V
)
###Class membership probabilities
lcpars[["pi_values"]] = apollo_mnl(mnl_settings, functionality="raw")
return(lcpars)
}
###busca pelos dados de entrada necessarios
apollo_inputs = apollo_validateInputs()
#cosntruindo a funcao de verossimilhanca
apollo_probabilities <- function(apollo_beta,apollo_inputs,functionality = 'estimate'){
###Attaches parameters and data so that variables can be referred by name
apollo_attach(apollo_beta, apollo_inputs)
on.exit(apollo_detach(apollo_beta, apollo_inputs))
###Create list for probabilities
P = list()
### Define settings for MNL model component
mnl_settings = list(
alternatives = c(auto=3, bus=2, metro=1),
avail = list(auto=1, bus=1, metro=1),
choiceVar = choice
)
### Compute class-specific utilities
V=list()
V[['auto']] = asc1 + b_ttauto[[1]]*tt_auto + b_coauto[[1]]*co_auto
V[['bus']] = asc2 + b_ttbus[[1]]*tt_bus_otp
V[['metro']] = asc3 + b_ttmetro[[1]]*tt_metro + b_cometro[[1]]*co_metro
###Calculating probabilities based on MNL function for class 1
mnl_settings$V = V
P[[1]] = apollo_mnl(mnl_settings, functionality)
### Compute class-specific regrets
R=list()
R[['auto']] = asc1 - log(1 + exp(b_ttbus[[2]]*tt_bus_otp-b_ttauto[[2]]*tt_auto)) -
log(1 + exp(b_ttmetro[[2]]*tt_metro-b_ttauto[[2]]*tt_auto)) - log(1 + exp(b_cometro[[2]]*co_metro-b_coauto[[2]]*co_auto))
R[['bus']] = asc2 - log(1 + exp(b_ttauto[[2]]*tt_auto-b_ttbus[[2]]*tt_bus_otp)) -
log(1 + exp(b_ttmetro[[2]]*tt_metro-b_ttbus[[2]]*tt_bus_otp))
R[['metro']] = asc3 - log(1 + exp(b_ttauto[[2]]*tt_auto-b_ttmetro[[2]]*tt_metro)) -
log(1 + exp(b_ttbus[[2]]*tt_bus_otp-b_ttmetro[[2]]*tt_metro)) - log(1 + exp(b_coauto[[2]]*co_auto-b_cometro[[2]]*co_metro))
###Calculating probabilities based on MNL function for class 2
mnl_settings$V = R
P[[2]] = apollo_mnl(mnl_settings, functionality)
###Calculating choice probabilities using class membership and conditional probabilities
lc_settings = list(inClassProb = P, classProb=pi_values)
P[["model"]] = apollo_lc(lc_settings, apollo_inputs, functionality)
P = apollo_avgInterDraws(P,apollo_inputs,functionality)
### Prepare and return outputs of function
P = apollo_prepareProb(P, apollo_inputs, functionality)
return(P)
}
Best regards,Model run using Apollo for R, version 0.1.1
www.ApolloChoiceModelling.com
Model name : mnl_lc18
Model description : usando acessos na escala logarit imica
Model run at : 2020-08-28 16:08:31
Estimation method : bfgs
Model diagnosis : successful convergence
Number of individuals : 6721
Number of observations : 6721
Number of cores used : 10
Number of inter-person draws : 300 (halton)
WARNING: Inter-person draws were used
without a panel data structure.
LL(start) : -7783.319
LL(0, whole model) : -2215132
LL(final, whole model) : -5097.986
Rho-square (0) : 0.9977
Adj.Rho-square (0) : 0.9977
AIC : 10237.97
BIC : 10381.04
LL(0,component_1) : -7383.773
LL(final,component_1) : -13989.86
LL(0,component_2) : -7383.773
LL(final,component_2) : -6798.264
Estimated parameters : 21
Time taken (hh:mm:ss) : 00:09:15.73
Iterations : 95
Min abs eigenvalue of hessian : 0.098924
Number of cores used : 1
Model without mixing
Gabriel
-
- Site Admin
- Posts: 1050
- Joined: 24 Apr 2020, 16:29
Re: Latent class with continuous random parameters: LL0 and rho2
Thanks, this should be fixed completely now in version 0.1.1