Binary Training Utility

A small library for training and evaluating binary classifiers with two selectable algorithm families.

Capabilities

Configurable training

• Training with algorithm "linear" on features [[0, 0], [0, 1], [1, 0], [1, 1]] and labels [0, 0, 0, 1] produces a fitted model whose label predictions match the labels on the same points when using predict_labels. @test
• Training with algorithm "ensemble" on features [[0, 0], [0, 1], [1, 0], [1, 1]] and labels [0, 1, 1, 0] produces a fitted model whose label predictions match [0, 1, 1, 0] on the same points when using predict_labels. @test
• Passing any algorithm string other than "linear" or "ensemble" raises a ValueError. @test

Predictions

• predict_probabilities returns an array shaped (n_samples, 2) where column 0 corresponds to label 0, column 1 corresponds to label 1, each row sums to 1.0, and the largest probability in each row corresponds to the label returned by predict_labels for the same input. @test

Evaluation

• evaluate_model trains using the chosen algorithm and returns an accuracy score between 0.0 and 1.0; on the [[0, 0], [0, 1], [1, 0], [1, 1]] and [0, 0, 0, 1] training data with identical test data it returns 1.0 when using the "linear" algorithm. @test

Implementation

@generates

API

from typing import Any, Literal
import numpy as np

Algorithm = Literal["linear", "ensemble"]

def train_model(features: np.ndarray, labels: np.ndarray, algorithm: Algorithm, random_state: int | None = None) -> Any:
    """Train a binary classifier using the selected algorithm and return the fitted model."""


def predict_labels(model: Any, features: np.ndarray) -> np.ndarray:
    """Return integer label predictions for each row in features."""


def predict_probabilities(model: Any, features: np.ndarray) -> np.ndarray:
    """Return probability estimates for class 0 then class 1, shaped (n_samples, 2)."""


def evaluate_model(
    train_features: np.ndarray,
    train_labels: np.ndarray,
    test_features: np.ndarray,
    test_labels: np.ndarray,
    algorithm: Algorithm,
    random_state: int | None = None,
) -> float:
    """Train on provided training data and report accuracy on provided test data."""

Dependencies { .dependencies }

scikit-learn { .dependency }

Provides supervised learning estimators, preprocessing utilities, and evaluation helpers.