lifelines 0.21.2

v0.21.2

on Python PyPI

0.21.2 - 2019-05-16

New features

• New regression model: PiecewiseExponentialRegressionFitter is available. See blog post here: https://dataorigami.net/blogs/napkin-folding/churn
• Regression models have a new method log_likelihood_ratio_test that computes, you guessed it, the log-likelihood ratio test. Previously this was an internal API that is being exposed.

API changes

• The default behavior of the predict method on non-parametric estimators (KaplanMeierFitter, etc.) has changed from (previous) linear interpolation to (new) return last value. Linear interpolation is still possible with the interpolate flag.
• removing _compute_likelihood_ratio_test on regression models. Use log_likelihood_ratio_test now.