Predict

Latest version: v3.1.17

Safety actively analyzes 687918 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 4 of 5

3.0.0

------------------

Added
~~~~~~~
- Move towards Python3 (Python2 not supported anymore!)
- Compatibility for Windows. Some small changes in functions, as some packages
behaviour differently under Windows. Also, adjusted sink and source paths
to use OS file separator.
- There is now for each feature group a parameter to determine whether the
features are extracted or not.

Changed
~~~~~~~
- Major: moved everything except the feature extraction to WORC!
- Scikit-image version, as older version was not compatible with numpy > 1.16
- Merge to Python3: Reduce function removed, imported from functools.
- Merge to Python3: repeat names of package when importing from same folder.
- Merge to numpy16: change in contour functions due to behavior of np.vstack
- Gabor features computed as last in texture features, as these take the longest.
- Use pydicom instead of dicom due to python3 merge.
- csv reader works different and impractical in python3. Use pandas instead
to read csv.
- Require PyRadiomics 2.1.2, as 2.2.0 seems to have a bug.

Fixed
~~~~~
- Bug in shape features or rather contour functions that specifically occured
on Windows.
- Except memory error for orientation features: make segmentation smaller.

2.1.3

------------------

Fixed
~~~~~
- Bug in SearchCVJoblib: estimator input removed.
- Bug in SearchCV: error when weights summed to zero in taking weighted average
and weighted std for the test scores.

2.1.2

------------------

Added
~~~~~~~
- If you have objects with missing features, put in NaN values.
- Several classifiers: LDA, QDA, GaussianNB and ComplementNB.
- CASH now also includes multiple machine learning model selection.
- Discrete uniform sampling
- average_precision_weighted scorer.

Changed
~~~~~~~
- Upgraded imputer to work with newest sklearn versions.
- Oversampling and SMOTE can now also be used in CASH optimization.
- Moved max_iter of all iterative classifiers and regressors to general config.
- Moved classifier construction to fitandscore to enable CASH.
- Surrogate for sklearn's check_scoring used to enable use of alternative
metrics.

Fixed
~~~~~
- Replaced several print statements to be Python3 compatible.
- SMOTE and Oversampling now only apply to the training set. Was previously
training plus validation.
- Scipy randint distribution was not compatible with the sklearn
ParameterSampler, so replaced them by a new discrete uniform distribution.
- Error catch when param_grid in CASH contains invalid values.

2.1.1

------------------

Added
~~~~~~~
- Relief feature selection.
- RankedSVM for multilabel classification.
- Multilabel workflow. All binary classifiers can now also be used
for multilabel classification true a sklearn OneVsRestClassifier
object.
- Multilabel performance metrics. Automatically used in plot_SVM
when using multilabel workflow.
- Full oversampling of underrepresented classes.
- log_uniform and exp_uniform distributions
- Volume now also computed when only providing a Nifti
- Single class Relief feature selection.
- MultiSVR ensembling.
- Start on F1_average metric.


Changed
~~~~~~~
- Reduced memory usage by not passing fitted objects, but just the
(hyper)parameters.
- All metrics used in evaluation are now combined in a metrics
script in the classification folder.
- Cleaned crossval function for readability.
- SVM hyperparameters are adjusted.
- GroupSel now has a separate Hyperparameter for all texture groups.
- Plot_barchart also compatible with above hyperparameter change.
- Ranked images are now named with their rank, making it easier to read the
output.

Fixed
~~~~~
- Catch error when validation set is too small.
- Logging of cross validation now works properly by removing previous
log filehandles.
- Ordering of methods in the SearchCV preprocess function now
corresponds with fitandscore.
- Bugs for multiclass predict in SearchCV and in plot_SVM
- Backend used in matplotlib is now agg, hence usage on platforms
not supporting tkinter is now supported.
- Raise an error when multiple label matches for a single feature file are found.
- Plot_SVM: Y_score needs to be computed with predict_proba, not the decision_function.
- SearchCV, Ensemble, Predict: Average was taken over binary scores. The
average is now taken over the posteriors.
- Plotting posteriors in ranked_images now correctly prints the posterior. Was
previously converted to integer.
- Samples get a unique ID in the crossvalidation, so we always now which sample was used.
- Bug in usage of PCA that caused it to never be used, plus a attribute name error.

2.1.0

------------------


Added
~~~~~

- PREDICT Exceptions in general.
- A PREDICT ValueError catch when fitting the classifiers has failed and thus
output is missing.
- Feature imputation added in fitandscore. Made additions in scripts parsing
these fields
towards fitandscore.
- Logistic regression classifier.
- Ensemble performance computing in plot_SVM.
- Feature selection based on variance taking the mean of the features into
account.
- Added PCA as feature selection method.
- Added plot_feature_groups function. Note that it's not used yet by default in
the trainclassifier analysis
- Some more docstrings and cleaned some functions.
- Save the cv_iter object in the SearchCV so you can re-use the
training/validation splits.
- Ensembling based on Caruana et al. 2004 in the SearchCV object.
- plot_SVM function also outputs percentage of crossval iterations in which
patient got classified correctly.
- Temporary save setting for crossval function is now in general config.
- Fastr execution plugin type is now also an argument for the SearchCV object.


Changed
~~~~~~~

- LoG features are now computed per 2-D slice, as is done with most other
features as well.
- SVM parameters slack is now dependent on features.
- Removed normalization in calcfeatures: now in separate node in WORC.
- Moved load_data function from trainclassifier to file_io for broader use.
- Removed preprocessing. It is now part of the WORC Python package and thus a
step separated from the feature calculation.
- Updated ttest tool to comply with new way of inputting feature files.
- SVM Polynomial kernel now spans [1, 5] instead of [2, 5]
- Cross validation splitting is now based on patient IDs, hence all samples
from all patients are either in the training or the testing set.
- Removed the general config.ini file and parser. These were used to determine
the settings for the Joblib Parallel function. These are now in the general
config. These had to be passed to both SMOTE and the Gabor feature computation.


Fixed
~~~~~

- The load_data function in trainclassifier didn't work properly for multiple
modalities.
- LoG filter needs a minimum of 4 voxels in each direction. These are therefore
now extracted on the full image instead of o nly the ROI slices. Plus there
is a catch if the full image also does not meet these criteria.
- SMOTE was not working properly in the nocrossval function in the classification/crossval.py script.
- Added ravel of labels to nocrossval function, as it was missing.
- Conversion to numpy uint8 for vessel features, as skimage cannot filter certain float images.
- construct_classifier gives an error for SVM's if a kernel type is not recognized.
- Param grids in construct_crossval were using loc and scale incorrectly.
- NaN in phase images are replaced by zeros to prevent errors.
- In vessel features, when segmentation is small, there is not inner or outer edge.
PREDICT gave an error, now those features are just zeros.
- In shape_features, first check if (0x18, 0x50) is present in metadata before
computing volume.
- No more unnecessary use of lists in CalcFeatures and get_features for the image data.
- Convert all integer inputs in compute_CI to floats, as divisions were
previously uncorrectly rounded when providing integers.
- plot_ROC function now works properly.
- Everything in the confidence interval function is now cast to floats,
as divisions with integers and floats are rounded to integers which gave
wrong values for the CI.
- Confidence interval now has a catch when only a single iteration is provided.
- Lot of error catches now have PREDICT errors.

2.0.0

------------------

Added
~~~~~

- Vessel, Laplacian of Gaussian (LoG), phase, GLCM multislice and NGTDM
features.
- Groupwise feature selection for these features.
- Function to perform statistical tests among features from classes.
- Selectfrommodel feature selection.
- SGDR, RFR, ElasticNet, and LASSO estimators.
- Option to supply fixed splits to cross validation.
- Option to perform no cross validation, but just a fixed train and test split.
- SMOTE?
- After each cross validation, a temporary save of the results of that
iteration can optionally be made.
- A lot of explanations/helps to functions.
- Plotting function for regression functions.
- Function to apply student t-test to features.

Changed
~~~~~~~

- Parameters of all features computations are now parsed from the configuration
ini file instead of solely being in the functions.
- Joblib and Fastr SearchCV objects now in single file. Created a general
SearchCV object.


Fixed
~~~~~

- SVM kernel degree can now also become linear.
- Feature group selection was not working.
- Regression now uses a non-stratified split for cross validation, a stratified
split is not possible and thus gave errors.
- There was a bug that the coliage features where always computed, which
is now fixed.

Page 4 of 5

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.