@wlearn/sdk

Convenience barrel that re-exports all wlearn model classes, AutoML, ensemble methods, pipeline, and core utilities in a single import.

Part of wlearn (GitHub, all packages).

Install

npm install @wlearn/sdk

Usage

const { XGBModel, LinearModel, autoFit, Pipeline, accuracy } = require('@wlearn/sdk')

const model = await XGBModel.create({ task: 'classification' })
model.fit(X, y)
console.log('accuracy:', accuracy(y_test, model.predict(X_test)))
model.dispose()

What is included

Models:

| Export | Package | |--------|---------| | LinearModel | @wlearn/liblinear | | SVMModel | @wlearn/libsvm | | XGBModel | @wlearn/xgboost | | LGBModel | @wlearn/lightgbm | | KNNModel | @wlearn/nanoflann | | EBMModel | @wlearn/ebm | | TsetlinModel | @wlearn/tsetlin | | BARTModel | @wlearn/stochtree | | XLearnLR, XLearnFM, XLearnFFM | @wlearn/xlearn | | MLPModel, TabMModel, NAMModel | @wlearn/nn | | RFModel, loadRF | @wlearn/rf | | GAMModel, loadGAM | @wlearn/gam | | ClusterModel, silhouette, calinskiHarabasz, daviesBouldin, adjustedRand, loadCluster | @wlearn/cluster | | MitraModel, registerMitraLoaders | @wlearn/mitra (optional) |

All model classes above are unified wrappers built with createModelClass. They accept an optional task parameter ('classification' or 'regression') and auto-detect the task from labels if omitted. Split classes (XLearnFMClassifier, MLPClassifier, etc.) are also re-exported for backward compatibility.

AutoML and ensemble:

• autoFit from @wlearn/automl
• VotingEnsemble, StackingEnsemble, BaggedEstimator from @wlearn/ensemble

Core utilities:

• Pipeline, load, loadSync, register
• encodeBundle, decodeBundle, validateBundle
• normalizeX, normalizeY
• StandardScaler, MinMaxScaler, Preprocessor
• accuracy, r2Score, f1Score, logLoss, rocAuc, and other metrics
• kFold, stratifiedKFold, trainTestSplit, crossValScore

Caveats

• Node/scripting only. Browser users should import individual packages so bundlers can tree-shake unused WASM binaries. The SDK pulls in all WASM modules (~30 MB total).
• @wlearn/mitra is an optional peer dependency (requires onnxruntime-node or onnxruntime-web). If installed, its exports are available; otherwise they are undefined.
• May lag behind individual model package releases.

License

Apache-2.0