scientist

Machine Learning library for Rust

Objective

Create Machine Learning abstractions to facilitate ML pipelines.

This package does not use interop with C or Python, it is meant to be written 100% in Rust.

crate - latest: 0.1.1

[dependencies]
scientist = "0.1.1"

Avaiable on this package

Linear Models

- Linear Regression ( Single dependant variable )
- Linear Classification ( Single dependant variable )

Usage

LinearRegressor

    extern crate scientist;
    use scientist::models::linear::{LinearRegression};

       let x_train : Vec<Vec<f64>> = [[1.1].to_vec(), [1.3].to_vec(), [1.5].to_vec(), [2.0].to_vec(), [2.2].to_vec(), [2.9].to_vec(), [3.0].to_vec()].to_vec();
       let y_train : Vec<f64> = [39343., 46205., 37731., 43525., 39891., 56642., 60150.].to_vec();
   
       let mut model : LinearRegression = LinearRegression::new();
   
       model.fit(&x_train, &y_train);
   
       let preds = m_model.predict(&[[1.5].to_vec()].to_vec());
   
       println!("Prediction {:?}", preds); // Prediction: $41434.737394958

TBD

RandomForest

- RandomForest Regression
- RandomForest Classification

Reinforcement Algorithms

- UCB
- Thompson Sampling

XGBoost

- XGBoost Regression
- XGBoost Classification