Scikit-learn's Fit/Predict pattern, ported to Go

What it does

GoLearn provides machine learning primitives for Go, wrapping tabular data into Instances objects that support matrix-like operations. You pass those instances to estimators, then evaluate results with built-in helpers for cross-validation, train-test splitting, and confusion matrices. The README walks through a KNN classifier on the Iris dataset, suggesting the library targets standard supervised learning workflows.

The interesting bit

The project’s main bet is familiarity: it mirrors Python’s scikit-learn Fit/Predict API so you can swap estimators with minimal friction. It also claims “simplicity, paired with customisability” as its goal, though the README is light on exactly which knobs are exposed.

Key highlights

• Mirrors a scikit-learn-style Fit/Predict estimator interface
• Loads CSV data into Instances structures with header preservation
• Includes evaluation helpers like confusion matrices and train-test splitting
• Ships with practical examples for KNN, trees, and instance manipulation
• Documentation available in English, Simplified Chinese, and Traditional Chinese

Caveats

• The README only surfaces KNN and tree examples, so the full algorithm roster is unclear
• “Customisability” is promised as a goal, but the sources do not detail how deep the tuning goes

Verdict

Worth a look if you want scikit-learn-style ergonomics in Go. Give it a pass if you need a broad algorithm catalog; the README only demonstrates KNN and trees.