|
2 | 2 | History |
3 | 3 | ======= |
4 | 4 |
|
5 | | -0.1.1 (2023-??-??) |
| 5 | +0.1.7 (2024-06-13) |
| 6 | +------------------ |
| 7 | +* Little's test implemented in a new hole_characterization module |
| 8 | +* Documentation now includes an analysis section with a tutorial |
| 9 | +* Hole generators now provide reproducible outputs |
| 10 | + |
| 11 | +0.1.5 (2024-04-17) |
| 12 | +------------------ |
| 13 | + |
| 14 | +* CICD now relies on Node.js 20 |
| 15 | +* New tests for comparator.py and data.py |
| 16 | + |
| 17 | +0.1.4 (2024-04-15) |
| 18 | +------------------ |
| 19 | + |
| 20 | +* ImputerMean, ImputerMedian and ImputerMode have been merged into ImputerSimple |
| 21 | +* File preprocessing.py added with classes new MixteHGBM, BinTransformer, OneHotEncoderProjector and WrapperTransformer providing tools to manage mixed types data |
| 22 | +* Tutorial plot_tuto_categorical showcasing mixed type imputation |
| 23 | +* Titanic dataset added |
| 24 | +* accuracy metric implemented |
| 25 | +* metrics.py rationalized, and split with algebra.py |
| 26 | + |
| 27 | +0.1.3 (2024-03-07) |
| 28 | +------------------ |
| 29 | + |
| 30 | +* RPCA algorithms now start with a normalizing scaler |
| 31 | +* The EM algorithms now include a gradient projection step to be more robust to colinearity |
| 32 | +* The EM algorithm based on the Gaussian model is now initialized using a robust estimation of the covariance matrix |
| 33 | +* A bug in the EM algorithm has been patched: the normalizing matrix gamma was creating a sampling biais |
| 34 | +* Speed up of the EM algorithm likelihood maximization, using the conjugate gradient method |
| 35 | +* The ImputeRegressor class now handles the nans by `row` by default |
| 36 | +* The metric `frechet` was not correctly called and has been patched |
| 37 | +* The EM algorithm with VAR(p) now fills initial holes in order to avoid exponential explosions |
| 38 | + |
| 39 | +0.1.2 (2024-02-28) |
| 40 | +------------------ |
| 41 | + |
| 42 | +* RPCA Noisy now has separate fit and transform methods, allowing to impute efficiently new data without retraining |
| 43 | +* The class ImputerRPCA has been splitted between a class ImputerRpcaNoisy, which can fit then transform, and a class ImputerRpcaPcp which can only fit_transform |
| 44 | +* The class SoftImpute has been recoded to better fit the architecture, and is more tested |
| 45 | +* The class RPCANoisy now relies on sparse matrices for H, speeding it up for large instances |
| 46 | + |
| 47 | +0.1.1 (2023-11-03) |
6 | 48 | ------------------- |
7 | 49 |
|
8 | 50 | * Hotfix reference to tensorflow in the documentation, when it should be pytorch |
| 51 | +* Metrics KL forest has been removed from package |
| 52 | +* EM imputer made more robust to colinearity, and transform bug patched |
| 53 | +* CICD made faster with mamba and a quick test setting |
9 | 54 |
|
10 | 55 | 0.1.0 (2023-10-11) |
11 | 56 | ------------------- |
|
0 commit comments