xlogit v0.2.7

• Fixed bug introduced in v0.2.6, which skipped estimation of standard errors by default in Multinomial Logit
• To improve a bit speed of the Mixed Logit estimation, the standard errors of the initialization using a MultinomialLogit model were skipped.

xlogit v0.2.6

• Implementation of option to skip standard errors by @gc-axa
  Note: This version introduced a small bug that skipped by default the estimation of standard errors in MultinomialLogit models. This was fixed in v0.2.7, which I recommend using instead of v0.2.6.

xlogit v0.2.5

• Implement additive term in the utility function to facilitate the modeling of coefficients kept fixed during estimation
• Migrated from TravisCI to GitHub Actions

xlogit v0.2.4

• Fix in import from scipy.stats

xlogit v0.2.3

• Fixes #12

xlogit v0.2.2

• Fix small bug in predict (#11)

xlogit v0.2.1

• Fixed bug with prediction for panel data

xlogit v0.2.0

• 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 :)

xlogit v0.1.9

• Significant improvements in estimation speed for Mixed Logit thanks to @jhelvy's formulation for gradient and log-likelihood.

xlogit v0.1.8

This release includes the following features implemented by @crforsythe

• Numerical hessian
• Robust standard errors