Latent class model

[Ying Lu]

Hi Prof. Bierlaire,

I am going to include socio-demographic variables into class membership model. However, I read several posts in this group and found out that it is not available to estimate a latent class choice model (including socio-demographics) with considering panel effects (https://groups.google.com/forum/#!searchin/biogeme/latent$20class$20model%7Csort:date/biogeme/3fwwvCjeI7o/kfYYsIAiFAAJ). 

Could I kindly know if you have a beta version of Biogeme which is able to include socio-demographics when estimating latent class choice model? Thank you!

Kind regards,

Ying 

[Bierlaire Michel]

Dear Ying,

This feature has now been implemented. I should be able to release it soon. 

Meanwhile, you can always reorganize your data so that all the data for one individual are on the same row, and the observations across days on several columns. 

Michel

--
You received this message because you are subscribed to the Google Groups "Biogeme" group.
To unsubscribe from this group and stop receiving emails from it, send an email to biogeme+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/biogeme/4e0bd595-f6ee-4453-a889-537ad0949469%40googlegroups.com.

[Nughthoh Arfawi Kurdhi]

Dear Prof.Bierlaire

I am a new user of Biogeme. 07discreteMixture.py is the example syntax for estimating latent class model. 

I have the following two questions:

1.   How to calculate the size of each class? (the number (percentage) of respondents on each class)

2.   Could we indentify individually which respondents in each class?

Thank you very much.

Best regards,

Arfa

Pada Sabtu, 07 September 2019 17.09.28 UTC+2, Bierlaire Michel menulis:

Dear Ying,

This feature has now been implemented. I should be able to release it soon. 

Meanwhile, you can always reorganize your data so that all the data for one individual are on the same row, and the observations across days on several columns. 

Michel

On 7 Sep 2019, at 12:21, Ying Lu <luyi...@gmail.com> wrote:

Hi Prof. Bierlaire,

I am going to include socio-demographic variables into class membership model. However, I read several posts in this group and found out that it is not available to estimate a latent class choice model (including socio-demographics) with considering panel effects (https://groups.google.com/forum/#!searchin/biogeme/latent$20class$20model%7Csort:date/biogeme/3fwwvCjeI7o/kfYYsIAiFAAJ). 

Could I kindly know if you have a beta version of Biogeme which is able to include socio-demographics when estimating latent class choice model? Thank you!

Kind regards,

Ying 

--
You received this message because you are subscribed to the Google Groups "Biogeme" group.

To unsubscribe from this group and stop receiving emails from it, send an email to bio...@googlegroups.com.

[Michel Bierlaire]

On 13 Sep 2019, at 11:53, Nughthoh Arfawi Kurdhi <ar...@mipa.uns.ac.id> wrote:

Dear Prof.Bierlaire

I am a new user of Biogeme. 07discreteMixture.py is the example syntax for estimating latent class model. 

I have the following two questions:

1.   How to calculate the size of each class? (the number (percentage) of respondents on each class)

If you sum the class membership probability over the sample, you’ll have the expected number of people in each class. 

Divide by the total number to have the percentage.

2.   Could we indentify individually which respondents in each class?

No. Only the probability. This is why the class is latent.

Thank you very much.

Best regards,

Arfa

Pada Sabtu, 07 September 2019 17.09.28 UTC+2, Bierlaire Michel menulis:

Dear Ying,

This feature has now been implemented. I should be able to release it soon. 

Meanwhile, you can always reorganize your data so that all the data for one individual are on the same row, and the observations across days on several columns. 

Michel

On 7 Sep 2019, at 12:21, Ying Lu <luyi...@gmail.com> wrote:

Hi Prof. Bierlaire,

I am going to include socio-demographic variables into class membership model. However, I read several posts in this group and found out that it is not available to estimate a latent class choice model (including socio-demographics) with considering panel effects (https://groups.google.com/forum/#!searchin/biogeme/latent$20class$20model%7Csort:date/biogeme/3fwwvCjeI7o/kfYYsIAiFAAJ). 

Could I kindly know if you have a beta version of Biogeme which is able to include socio-demographics when estimating latent class choice model? Thank you!

Kind regards,

Ying 

--
You received this message because you are subscribed to the Google Groups "Biogeme" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bio...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/biogeme/4e0bd595-f6ee-4453-a889-537ad0949469%40googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "Biogeme" group.

To unsubscribe from this group and stop receiving emails from it, send an email to biogeme+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/biogeme/912f59db-709f-4835-bbf5-b388b173ad36%40googlegroups.com.

[Michel Bierlaire]

If the class membership is fixed, this ius easy: 

class 1 : 25.1 % of the population

class 2: 74.9 % of the population. 

On 17 Sep 2019, at 12:25, Nughthoh Arfawi Kurdhi <ar...@mipa.uns.ac.id> wrote:

Dear Prof. Michel Bierlaire

Thank you for the answers. In the example 07discreteMixture.py the class membership model is

W_OTHER

0.749

then we have

probClass1 = 0.251
probClass2 = 0.749

I still do not understand how to use this information to sum the class membership probability over the sample in order to ge the size of each class.

Thank you very much in advance.

Best regards

Arfa

[Rim]

Hello Prof. Bierlaire, and other members,

I have an additional question about the calculation of the membership probability.

If we are in a case similar to the one presented in 16panelDiscreteSocioEco  (but simpler with no random coefficients). So, we have sociodemographic variables influencing the probability of being in class 0 or class1.

How could we calculate the probabilities?

Thank you in advance for any help,

Rim