Tabular ML

Tabular-ML is a Rust library I built primarily to practice using the Rust programming language and familiarizing myself with its package system. I chose a project around tabular ML because it's what I know best! The project implements core tabular machine learning algorithms, designed for learning, experimentation, and practical use on structured data. It provides a familiar API for users coming from Python's scikit-learn, but leverages Rust's performance and safety.

Features

• Linear Models

  • Linear Regression (Ordinary Least Squares)
  • Ridge Regression (L2 regularization)
  • Lasso Regression (L1 regularization)
  • Elastic Net (L1 + L2 regularization)
  • Logistic Regression (binary classification)
  • Support Vector Machine (SVM) for binary classification (multiple kernels)

• Clustering

  • K-Means
  • DBSCAN

• Dimensionality Reduction

  • Principal Component Analysis (PCA)
  • Linear Discriminant Analysis (LDA)
  • Truncated Singular Value Decomposition (TruncatedSVD)

• Preprocessing

  • StandardScaler (feature standardization)
  • Dataset utilities (train/test split, etc.)

• Metrics

  • Regression and classification metrics (R², MSE, accuracy, precision, recall, F1, confusion matrix, etc.)

Getting Started

Add to your Cargo.toml:

[dependencies]
nametbd = { path = "path/to/this/repo" }
ndarray = "0.16"

Example Usage

Linear Regression

use nametbd::{LinearRegression, Matrix, Vector};
use ndarray::array;

fn main() -> Result<(), String> {
    let x = array![[1.0], [2.0], [3.0], [4.0], [5.0]];
    let y = array![5.1, 6.9, 9.2, 11.1, 12.8];

    let mut model = LinearRegression::new();
    model.fit(&x, &y)?;
    let predictions = model.predict(&x)?;

    println!("Predictions: {:?}", predictions);
    Ok(())
}

K-Means Clustering

use nametbd::{KMeans, Matrix};
use ndarray::array;

let x = array![
    [1.0, 1.0],
    [1.5, 2.0],
    [3.0, 4.0],
    [5.0, 7.0],
    [3.5, 5.0],
    [4.5, 5.0]
];

let mut kmeans = KMeans::new(2).max_iter(100);
let labels = kmeans.fit_predict(&x).unwrap();
println!("Cluster labels: {:?}", labels);

Principal Component Analysis (PCA)

use nametbd::{PCA, Matrix};
use ndarray::array;

let x = array![
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0]
];

let mut pca = PCA::new().n_components(2);
let transformed = pca.fit_transform(&x).unwrap();
println!("PCA output: {:?}", transformed);

Examples

See the examples/ directory for:

• Linear regression, logistic regression, and regularized regression
• Clustering (KMeans, DBSCAN)
• Dimensionality reduction (PCA, LDA, TruncatedSVD)
• SVM classification

Run an example with:

cargo run --example simple_example

Design Goals

• Educational: Clear, readable implementations of core algorithms.
• Extensible: Modular design for adding new models and utilities.
• Safe and Fast: Leverages Rust's safety and performance.
• Familiar API: Inspired by scikit-learn for easy adoption.

Roadmap

Planned features include:

• Tree-based models (Decision Trees, Random Forests, Gradient Boosting)
• Naive Bayes classifiers
• Ensemble methods (Bagging, AdaBoost, Voting)
• Model selection and validation (cross-validation, grid search)