Fix breaking updates in sklearn

Merge pull request #120 from havakv/bug-fix

Bump version: 0.2.2 → 0.2.3

Bug fixes and dedicated data storage

Use py7zr for uncompressin kkbox data (#69)

Co-authored-by: Haavard Kvamme <haavard.kvamme@abelee.com>

Fix numba changes and update python and pytorch version

Update to python 3.8 and pytorch 1.5 (#36)

* Update to python 1.5

* Update setup.py

* Bump version: 0.2.0 → 0.2.1

Co-authored-by: havakv <haavard.kvamme@gmail..com>

Update to work with torchtuples v0.2.0

Release notes

Features

• Administrative Brier score and Binomial log-likelihood for evaluation of data sets with administrative censoring.

• BCESurv which is a method that disregards censoring and does not enforce monotone survival functions. It is meant to represent a set of binary classifiers that disregards censored observations.

• Improved kkbox data sets with administrative censoring times and more covariates.

• sac_admin5 simulated data set with administrative censoring.

• More simulations studies with covariate dependent censoring times and administrative censoring.

Changes

• Updated to work with torchtuples v.0.2.0

• CoxPH now use a regular data set, instead of the durations sorted. The old method is renamed CoxPHSorted but will be removed.

PyPI package

Minor bug fixes and release to PyPI.

v0.1.0

Release notes v0.1.0

These note mainly focus on the changes to existing code and not new functionality.

Evaluation criteria

• Fixed wrong index in IPCW.
• EvalSurv now has a steps argument determining how the survival curve should behave between estimated times.
  Previously set to 'pre', but now 'post' is default.
  This will affect the concordance for the discrete-time methods the most. Set ev.step = 'pre' to obtain old results.
  Or use some reasonable interpolation scheme.
• Moved pycox.evaluation.utils to pycox.utils.
• Replaced the binomial log-likelihood mbll with the negative binomial log-likelihood nbll. I.e. only the sign is different.

Models

• Replaced predict_survival_function with predict_surv and predict_surv_df.
• More stable version of CoxCC and CoxTime loss for single control.
• Restructured the locations of the Cox models.

Preprocessing

• Added quantiles discretization for methods.