Missing-Value Imputation
Impute correlated numeric features in browser and Node.js with the @kanaries/ml JavaScript IterativeImputer implementation.
View as MarkdownMissing values can break estimators or bias a complete-case analysis. The Impute module supplies transformer-style preprocessing that fits only on training data and can be placed inside repeatable JS pipelines.
Use IterativeImputer when features predict one another and simple column means discard too much structure.
Feature Hashing and Dictionary Vectorization
Use HashingVectorizer, DictVectorizer, and FeatureHasher for bounded-memory JavaScript and TypeScript feature extraction.
IterativeImputer
Estimate each missing numeric feature from the others using the @kanaries/ml JavaScript IterativeImputer implementation in browser or Node.js.