nyoka

• Added support for xgboost till version 1.7.6
• Added support for statsmodels till version 0.14.0
• Added support for scikit-learn till version 1.3.0

• Added a custom exporter to convert pipeline with XGBoost models to PMML.

• Added algorithmName="randomForest" attribute as part of MiningModel element for Random Forest models (PR - #56)
• Added the derived fields defined in the TransformationDictionary section of the MiningModel as MiningFields within each of the Segment models for Random Forest models (PR - #57)

• Added support for xgboost 1.x.x version (till 1.5.2) #53

Performance Improvement

• Improved performance of xgboost exporter to a greater extend (PR - #50)

Bug fix

• #39 - For LightGBM exporter
• Tree node split threshold is now converted from float64 to float32 for scikit-learn's tree based models.

Major Changes

• Removed customized elements from PMML schema. Now, Nyoka is completely Official PMML 4.4.1 schema compliant.
• Dropped support for Keras and Retinanet exporter as these two exporters were using customized DeepNetwork element.

CICD

• Usage of Travis-CI is discontinued. Instead of that Github Actions is used

Minor Update

• Added support for PMML schema 4.4.1

Minor update

• Added support for the latest version of scikit-learn (<=0.23.1)

Bug fix

• Removed Python version constraint from setup.py. Now user can install nyoka when python version is >= 3.6

Added exporter for TrendMiner's fingerprint

• This can be imported as from nyoka.custom.trendminer import FingerprintToPmml

Minor Update

• Added support for VARMAX from statsmodels library (multi-variate time series model)
• ArimaToPMML is deprecated and StatsmodelsToPmml is introduced
• conf_int parameter is added to StatsmodelsToPmml which will generate OutputField for lower and upper bound of confidence interval
• Added support for ARIMA in state space form from statsmodels library.

Structural change

• Added nyoka.base.enums.py which contains enum for all enumerations present in PMML schema
• Added example jupyter notebook for VARMAX

Schema update

• Added InterceptVector, PredictedStateCovarianceMatrix and SelectedStateCovarianceMatrix as child element and observationVariance as attribute of StateSpaceModel element
• Two new result-feature confidenceIntervalUpper and confidenceIntervalLower are added to schema

Minor Update

• model_name and description parameters are added to all the exporters.
• Updated PMML schema for KalmanState element
• Updated Arima exporter to export outputfield for predictionInterval with feature type "standardError"

Bug fixes

• Now the endog for ARIMA/SARIMAX can be any iterable
• Timestamp value in Header is updated from utcnow() to now()

Bug Fix

• Bug fix for ArimaToPMML class (AttributeError when DataFrame is used for training)

Major Update

1. Added exporter for Object Dectection Model (RetinaNet)
2. Added extra parameter to Keras exporter to add python script export functionality.
3. Reduced list of parameters in ARIMA exporter

Bug Fix

1. Bug fix for Seasonal ARIMA (Residual for Seasonal Component)
2. Bug fix for AnomalyDetectionModel (Calculation of the threshold)
3. Bug fix for NeuralNetwork models

Minor update

• PMML representaion updated
  • AnomalyDetectionModel is used for IsolationForest instead of MiningModel
  • stddev as a build-in function
  • Removed threshold from OutputField

Bug fix

• Changed operator in SimplePredicate for LightGBM
• Changed default wordSeparatorCharacterRE in TextIndex from (?u)\b\w\w+\b to \s+
• Corrected invalid constantTerm, TimeValue in ARIMA.
• Added RMSE, Residuals to ARIMA

Bug fix

• Output dataType is updated to have same data type as the classes have.
• batchNormalizationCenter is added to LayerParameters
• Removed unused PMML elements generated in case of LabelBinarizer and OneHotEncoder
• feature_importances_ not found in IsolationForest

Minor Update

• Added pre-processing Transformer Lag

Bug Fix

• OneHotEncoder has not attribute _active_features_
• Parameters for ArimaToPmml and ExponentialSmoothingToPmml have made optional. Now users need to pass only result_objs to the exporter. No need to pass time_series_data and model_obj

Change in PMML representation

• Logistic Regression
• Ridge Classifier
• SGD Classifier
• Linear SVC
• Linear Discriminant Analysis

Bug fix

• Linear SVC does not support probability
• KMeans Comparison Measure
• PCA, Binarizer, Gradient Boosting Classifier, Gradient Boosting Regressor change in precision
• CountVectorizer generates unsupported bytes.
• Removed duplicate OutputField

Bug Fix

• Parameters change in KerasToPmml init()

Bug Fix

• Bug fix for Keras Exporter