1.2.0

What's Changed

• fix(Calibration): propagate pos_label throughout by @almostintuitive in #206
• feature(Utils): Export ParsableEnum. by @szemyd in #205
• feature(Library): Added imbalance metrics. by @szemyd in #207
• feature(Analysis): Random model updated to have a smoothing factor. by @szemyd in #208
• fix(ScoreCard): ScoreCard indexing should return a copy not the actual object. by @szemyd in #209
• fix(Metric): Imbalance warning by @szemyd in #210
• feature(Evaluate): Random model has a shuffled chunk version. by @szemyd in #213
• fix(Evaluate): Random chunking now supports length that are not wholly divisible by chunk_size. by @szemyd in #214
• feature(Metrics): added avg_precision metric by @almostintuitive in #212
• fix(Evaluation): Benchmark metrics were switched around. by @szemyd in #216
• chore(Analysis): Matplotlib and Statsmodel dependency optional now. by @szemyd in #215
• feature(Metrics): pred_y_imbalance_ratio is now the absolute difference between the ratio of y and pred's imbalance by @almostintuitive in #211
• chore(Exporting): Export all metrics to be accessed from krisi.library. by @szemyd in #217
• feature(Data): Shuffling in chunks updated to work with copy_on_write = True. by @szemyd in #218
• fix(Metrics): disable rolling metrics for roc_auc by @almostintuitive in #219
• feature(Modeling): Random model shuffling now accpets float as chunk_size. by @szemyd in #220
• fix(Utils): shuffle_df_in_chunks should only shuffle indices, not the whole dataframe by @almostintuitive in #222
• feature(Evaluation): New SampleTypes - Validation Sample. by @szemyd in #221
• feature(Util): Handle datetime index in shuffling in chunks. by @szemyd in #223
• feature(Utils): Combinatorial shuffling. by @szemyd in #224

Full Changelog: 1.1.1...1.2.0

1.1.1

What's Changed

• fix(Library): put back rmse to default metrics by @almostintuitive in #200
• fix(Dev): Raise error when raise_exception passed in. by @szemyd in #201
• fix(Group): Group is now handled separately by @szemyd in #202
• feature(Utils): Exposing cross-package functions. by @szemyd in #203
• feature(Library): added distinction between balanced vs imbalanced metrics by @almostintuitive in #204

Full Changelog: 1.1.0...1.1.1

1.1.0

What's Changed

• fix(Evaluation): Brier score default metric fixed when columns not set correctly. Updated in and out of sample scoring function. Added test. by @szemyd in #168
• feature(Metrics): added binary vs multiclass metrics, get default metrics automatically by @almostintuitive in #174
• fix(Metrics): wrap roc_auc score so it works when doing binary classification by @almostintuitive in #175
• fix(Metrics): only used weighted and macro roc_auc score for multi-label by @almostintuitive in #176
• fix(Utils): correctly recognize dataset types by @almostintuitive in #179
• feature(Group): Metric consistency over time. by @szemyd in #180
• feature(Evaluation): Added support for sample_weights. by @szemyd in #182
• fix(Printing): Scorecard get's printed multiple times to console. by @szemyd in #183
• feature(Metrics): added ncdg, changed default metrics slightly by @almostintuitive in #181
• fix(Evaluation): Sample weight rolling window fix. by @szemyd in #184
• fix(Report): Callibration plot. by @szemyd in #185
• fix(Evaluation): Rolling evaluation shouldn't assume the name of the arguments. by @szemyd in #186
• fix(Library): use ovr instead of ovo for roc_auc by @almostintuitive in #187
• feature(Library): Added Bennet's et. al S-Score classification metric. by @szemyd in #188
• feature(Report): Calibration plot in console & took out rmse from default metrics. by @szemyd in #190
• feature(Library): added weighted metrics to binary presets by @almostintuitive in #191
• feature(State): Added global state that allows us to throw errors in try except when in RunType.test. Fixed S-Curve by @szemyd in #193
• feature(Checks): check if sample_weight has the same length as target by @almostintuitive in #194
• fix(Tests): don't use matplotlib if not installed by @almostintuitive in #196
• feature(Evaluation): Benchmarking metrics to a random model. by @szemyd in #195
• feature(Test): Don't show plots when we are testing. by @szemyd in #197
• fix(Data): Syntethetic data generation. by @szemyd in #198

Full Changelog: 1.0.1...1.1.0

1.0.1

What's Changed

• fix(ScoreCard): ipython should only be imported if notebook is available. by @szemyd in #166

Full Changelog: 1.0.0...1.0.1

1.0.0

What's Changed

• feature(ScoreCard): Saving now handles overwriting based on model_name by @szemyd in #160
• fix(ScoreCard): We should clean the directory first in tests and misspelling. by @szemyd in #161
• feature(Scoring): Handle probabilities and multiclass classification. by @szemyd in #162
• feature(Report): Calibration plot and many fixes. by @szemyd in #164

Full Changelog: 0.1.4...1.0.0

0.1.4

What's Changed

• feature(Analyse): Exposed correlation over time metrics in analyze module. by @szemyd in #154
• feature(Docs): Rewamped the README with new examples, changed wording and new illustrations by @szemyd in #155
• feature(Docs): Updated walkthrough by @szemyd in #156

Full Changelog: 0.1.3...0.1.4

0.1.3

What's Changed

• feature(Report): Better pdf reporting, direct_save, direct_one_subplot option for generate_report, refractored code, new examples. by @szemyd in #152

Full Changelog: 0.1.2...0.1.3

0.1.2

What's Changed

• feature(Report): Calling the object in the terminal will output a pretty printed version and the orginal python __repr__ for ScoreCard and Metric. by @szemyd in #150
• feature(Analysis): A new analysis module that explores features on a rolling basis. EDA with finer insights by @szemyd in #151

Full Changelog: 0.1.1...0.1.2

0.1.1

What's Changed

• feature(Docs): Added example gallery and updated preferences. by @szemyd in #138
• fix(Docs): Updated readme with the correct example link. by @szemyd in #139
• feature(Docs): Updated docs to include better API. by @szemyd in #140
• feature(CI): added bump-version workflow by @almostintuitive in #142
• feature(Metric): Metrics can be grouped now to make precalculating possible. by @szemyd in #143
• feature(Rolling): Upgraded the rolling function and unified functions. We are using pd.rolling now. by @szemyd in #145
• chore(Modeling): Modeling removed from krisi, df displayed in notebook with ipython.display. by @szemyd in #146

Full Changelog: 0.1.0...0.1.1

0.1.0

What's Changed

• feature(Report): Vizualise util function for predictions and targets by @szemyd in #113
• feature(Report): Added clearer vizualisation of prediction results. by @szemyd in #114
• feature(Docs): Badges as styled for external use on our website. by @szemyd in #115
• feature(Vizualisation): Seperate y plot, fixed index issue. by @szemyd in #117
• feature(Vizualisation): Unified hover, renamed defaults and allow for unwrapped modes. by @szemyd in #118
• feature(Report): You can now pass in a pd.Series to plot_y_predicions. by @szemyd in #120
• feature(Dataset): Added utils to check consistency of a dataset. by @szemyd in #119
• feature(Scorecard): Slicing Scorecard and printing multiple versions of the scorecard by @szemyd in #124
• feature(Docs): Added explanation of the print function and fixed minor remaining error with the previous commit. by @szemyd in #125
• feature(Report): Compare function now receives dataframe boolean to print a df rather than a formatted string. by @szemyd in #129
• feature(Report): Compare method now handles unspecified metrics, sort_metric_key now optional, if not provided it will be the first element of metrics_to_display by @szemyd in #130
• fix(Report): Compare printing bold didn't work in jupyter notebook. by @szemyd in #131
• chore(Report): Cleaned up argument names of compare and added docstrings. by @szemyd in #132
• chore(Docs): Updated readme with Dream Faster suite. by @szemyd in #133
• chore(Docs): Renamed reference to fold-wrappers by @szemyd in #134
• fix(Docs): updated tests badge in README by @almostintuitive in #136

Full Changelog: 0.0.8...0.1.0