anofox_ml

51 downloads this week 3 stars Extension descriptor

Machine learning in SQL - 188 scikit-learn-compatible estimators (trees, boosting, linear models, GLMs, SVMs, kNN, naive Bayes, MLPs, clustering, decomposition, preprocessing, text vectorizers, meta-estimators, conformal prediction), fitted per group with an aggregate and stored as BLOBs, plus metrics, cross-validation, tuning and drift detection.

Compatibility Matrix

This table covers all DuckDB releases that have not reached end-of-life, including LTS releases, and shows the 2 most recent patch releases of each. Older patch releases are not listed, but the extension may be available for them as well.

Installing and Loading

INSTALL anofox_ml FROM community;
LOAD anofox_ml;

Example

CREATE TABLE train AS
  SELECT (i % 2)::DOUBLE AS y, [i::DOUBLE, (i % 2) * 3.0 + (i % 5) * 0.1] AS x
  FROM range(100) t(i);
-- fit a model in an aggregate; it is an ordinary BLOB value
CREATE TABLE models AS
  SELECT ml_fit_agg('RandomForestClassifier', y, x, '{"n_estimators": 50, "seed": 42}') AS model
  FROM train;
SELECT ml_predict(model, [7.0, 3.2]) AS label,
       ml_predict_proba(model, [7.0, 3.2]) AS proba
FROM models;

About anofox_ml

anofox_ml brings the anofox-ml Rust library into SQL. ml_fit_agg(kind, y, x [, params]) fits any of 188 estimators (see ml_estimators()), one model per GROUP BY group in parallel; the result is a versioned, compressed BLOB you can keep in a table or Parquet file and use with ml_predict, ml_predict_proba, ml_transform, ml_predict_interval and the other model functions in any later query. It also has metric aggregates, cross-validation splitters, grid / random / halving search (ml_tune_agg), permutation importance, recursive feature elimination and drift detection. Documentation: https://github.com/DataZooDE/anofox-ml

Added Functions

Overloaded Functions

This extension does not add any function overloads.

Added Types

This extension does not add any types.

Added Settings