Blyx

Classification

Metrics library · 0.1.1

Measure threshold decisions, false alarms, missed positives and their costs.

Install with Conda

Create an environment containing Classification 0.1.1, its Blyx source and a runnable example. Run the second command with Blyx 0.2.1, Rust and a host linker installed.

conda create -y --prefix ./blyx-env --override-channels -c https://rahulchaube1.github.io/blyx-packages blyx-classification=0.1.1=0
blyxc native-run ./blyx-env/share/blyx/classification/main.blyx

To use this package in your application, copy share/blyx/classification/vendor/classification from the environment into your project's vendor directory.

Install the full collection with Homebrew or Scoop

Import into a Blyx project

Install Blyx 0.2.1, Rust 1.85 or newer and a host linker. From the directory containing your entry file, clone this tagged release.

git clone --depth 1 --branch v0.1.1 https://github.com/Rahulchaube1/blyx-classification.git vendor/classification

Save this as main.blyx beside vendor and run blyxc native-run main.blyx.

use vendor::classification::classification;

fn main() {
    let table = class_confusion([1, 0, 1, 0], [0.9, 0.6, 0.4, 0.1], 0.5);
    println!(class_precision(at(table, 0), at(table, 1)));
    println!(class_recall(at(table, 0), at(table, 3)));
    println!(class_cost(at(table, 1), at(table, 3), 1.0, 5.0));
}

Expected output

0.5
0.5
6

API and input requirements

Inputs are equal-length vectors of binary labels and finite probabilities. A score equal to the threshold counts as positive. An undefined precision, recall or F1 has value zero.

APIUse
class_confusion(labels, scores, threshold)Return true positives, false positives, true negatives and false negatives in that order.
class_precision, class_recall, class_f1Calculate scalar metrics from the confusion counts.
class_cost(fp, missed, fp_cost, fn_cost)Compute total weighted cost. Divide by observation count for cost per row.

Develop and contribute

Open the tagged source to read the implementation and examples. Include the smallest reproducible Blyx program, compiler version and expected result when reporting an issue.