Binary prediction with time serial and dummys

Question

I have this data set.

I have 4 individuos, Y binary dependent variable.

I want calculate (prediction) the probability of sale of the car d (?%), in the years 2001, 2002, 2003.

The variables macroeconomics is for each city, but for some the values it is the same, I want than the model capture this relation and can predict well a car in a city with few train data using the train date of other city with the same macroeconomic variable o very similar. I have variables of the car, and they is atemporal, it have the same value for all years.

I saw a Bayesian approach to solve this problem, but I think that the an easier solution can be found.

Answer 1

How about using a generalized linear mixed effect model. What I mean by that is you probably want to fit a longitudinal logistic regression. You have a single binary outcome variable which means logistic regression, further you have dependency where order matters (i.e. your time variable).

I strongly feel that using a GLMM will do what you want it to do.