- WTP models for Mixed Logit and Multinomial Logit implemented with the support of jhelvy and crforsythe - Optimized probabilty product across panels - Optimized `utils.wide_to_long` - Added AI-generated logo :)
0.1.9
- Significant improvements in estimation speed for Mixed Logit thanks to jhelvy's formulation for gradient and log-likelihood.
0.1.8
This release includes the following features implemented by crforsythe - Numerical hessian - Robust standard errors
0.1.7
- Added ability to batch across samples (choice situations). This will significantly help when working with large datasets - Changed the mechanism to handle panels. Instead of adding extra dimensions, now the product for probabilities is done across panel ids. - Fixed bugs
0.1.6
- Fix avail parameter for panel data in Mixed Logit