ApolloChoiceModelling forum

Hi Stephane,

I am trying to specify a latent class model. But the class is not fixed for each individual over time. In other words, starting from one period, an individual's situation is changed, his status/group is changed. In model specification, I want to include not only individual demographic variables, but also time specific variablbes to indicate that class changes. Is that possible in Apollo?

Thanks,
Sheldon

Hi, sure, this is possible. What you need to do is to only have the call to apollo_panelProd after the averaging across classes, i.e. apollo_lc. So essentially, this is an intra-individual version of LC. The call to apollo_panelProd at the end will mean that you still recognise the repeated choice nature of the data when calculating the standard errors

just so say that there is a small bug preventing this from working at the moment. It's being fixed for version 0.2.8, the development version of which will be on our website in the coming days

stephanehess wrote: 02 Feb 2022, 22:22 Hi, sure, this is possible. What you need to do is to only have the call to apollo_panelProd after the averaging across classes, i.e. apollo_lc. So essentially, this is an intra-individual version of LC. The call to apollo_panelProd at the end will mean that you still recognise the repeated choice nature of the data when calculating the standard errors

Hi Stephane,

could you provide some examples about the intra-individual version of LC? in the manual of Apollo, I do not see this kind of explanation. in my latent class model, I assume that under different travel context, such as travel purpose, individual may belong to different classes which have different model specification. however, to this problem, I do not know how to solve with the class membership probability. in a inter-individual version, the probability is π n,s, representing individual n in class s. therefore, if it is an intra-individual version, should I set the probability of π n,s,t in choice task t?

Best,
Qiang

Hi

I don't have any such example files, but like I said, this would all be accommodated by moving the apollo_panelProd call to be after apollo_lc. It's a very simply change for you to make

For the class allocation model, you could then include covariates that vary across observations in the class allocation model too

Stephane