I have two high dimensional vectors (~150 dims), assumed to be sampled from multinomial distributions. How can I test whether the two are from the same multinomial distribution or different ones? (Especially if the samples are small.)
Asked
Active
Viewed 38 times
2
kjetil b halvorsen
- 63,378
- 26
- 142
- 467
user294869
- 31
- 2
-
Hello, I guess with this high dimensions, it is impossible to fit multinomial model and use a Wald/score/likelihood ratio test.. I think it is possible to use a LASSO regression and perform test with it, but I'm not familiar with that. This [question](https://stats.stackexchange.com/questions/241082/testing-for-coefficients-significance-in-lasso-logistic-regression) deals with that for logistic regression. An easy option could be to go Bayesian, use a non flat prior for the probabilities of the models and do Bayesian model selection... – Pohoua Nov 16 '20 at 00:27