1.0.3

• Do not remove groups property from symbol table

1.0.2

• Fix KNN and Hot Deck imputer reset donor samples

1.0.1

• Fix AdaMax optimizer when tensor extension loaded
• Prevent certain specification false negatives
• Add extension minimum version specification

1.0.0

• No changes

1.0.0-rc1

• Added Token Hashing Vectorizer transformer
• Added Word Stemmer tokenizer from Extras
• Remove HTML Stripper and Whitespace Remover transformers
• Rename steps() method to losses()
• Steps() now returns iterable progress table w/ header
• Remove rules() method on CART
• Removed results() and best() methods from Grid Search
• Change string representation of NAN to match PHP
• Added extra whitespace pattern to Regex Filter

1.0.0-beta2

• Interval Discretizer now uses variable width histograms
• Added TF-IDF sublinear TF scaling and document length normalization
• Dataset filterByColumn() is now filter()
• Added Lambda Function transformer from Extras
• Rename Dataset column methods to feature
• Added Dataset general sort() using callback
• Confusion Matrix classes no longer selectable
• Remove Recursive Feature Eliminator transformer
• Metric range() now returns a Tuple object

1.0.0-beta1

• Added variance smoothing to Gaussian NB, Mixture, and MLE
• Added MAD smoothing to Robust Z Score
• Added Writable extractor interface
• NDJSON and CSV extractors are now Writable
• Added SQL Table dataset extractor
• Changed Word Count Vectorizer DF constraints to proportions
• Change order of Naive Bayes hyper-parameters
• Persisters use RBX serializer by default
• Removed previously deprecated portions of the API
• Removed Embedder interface and namespace
• Change Robust Z Score alpha parameter name to beta
• Hold Out validator does not randomize by default
• Move Redis DB persister to extras package
• Remove Loda estimate bins static method
• Change Grid Search base estimator param name to class
• Remove Dataset cast to string preview
• Add Error Analysis error standard deviation and drop midrange
• Naive Bayes Laplace smoothing no longer effects priors
• Nearest Neighbors distance weighting off by default
• Promoted the Other namespace
• Moved Flysystem persister to the Extras package
• Change order of Loda hyper-parameters
• Persistent Model now accepts an optional serializer
• Persisters no longer interact directly with Persistables
• Remove Wrapper interface
• RBX serializer now accepts base Gzip parameter
• Gzip serializer no longer accepts base serializer
• Changed Gzip default compression level from 1 to 6
• Changed RBX default compression level from 9 to 6
• Do not persist training progress information
• Change underscores in Report property names to spaces
• Add saveTo() method to Encoding object
• Add Dataset exportTo() method
• Pipeline and Committee Machine are no longer Verbose
• Remove K Best feature selector (special case of RFE)
• Changed Error Analysis metrics
• Remove threat score from Multiclass Breakdown
• Rename Labels Are Missing exception
• Feature importances are no longer normalized
• Optimized CART binary categorical node splitting
• Interval Discretizer outputs numeric string categories
• Renamed Random Hot Deck Imputer
• Changed order of decision tree hyper-parameters

0.4.1

• Optimized CART node splitting for low variance continuous features
• Fixed RBX serializer string representation
• Prevent overwrites when instantiating Unlabeled from iterator

0.4.0

• Added Truncated SVD transformer
• Added Rubix Object File (RBX) format serializer
• Added class revision() method to the Persistable interface
• Added custom class revision mismatch exception
• Add Boolean Converter transformer
• Deprecated Igbinary serializer and move to Extras package
• Deprecate explainedVar() and noiseVar() methods on PCA and LDA
• Added missing extension specification and exception

0.3.2

• Fix t-SNE momentum gain bus error when using Tensor extension
• Optimize t-SNE matrix instantiation
• Refactor single sample inference methods
• Update the docs site