Dendritic

Dendrite is a general purpose supervised/un-supervised machine learning library written for the rust ecosystem. It contains the required data structures & algorithms needed for general machine learning. It acts as core library with packages for predictive data modeling.

Disclaimer

The dendritic project is a toy machine learning library built for learning and research purposes. It is not advised by the maintainer to use this library as a production ready machine learning library. This is a project that is still very much a work in progress.

Published Crates

Building The Dendritic Packages

Dendritic is made up of multiple indepedent packages that can be built separatley. To install a package, add the following to your Cargo.toml file.

[dependencies]
dendritic = { version = "<LATEST_VERSION>", features = ["bundled"] }

Example IRIS Flowers Prediction

Down below is an example of using a multi class logstic regression model on the well known iris flowers dataset. For more examples, refer to the dendritic-models/src/main.rs file.

use dendritic_datasets::iris::*;
use dendritic_regression::logistic::*;
use dendritic_metrics::loss::*;
use dendritic_metrics::activations::*;
use dendritic_preprocessing::encoding::*;


fn main() {

    // load data
    let data_path = "../../datasets/data/iris.parquet";
    let (x_train, y_train) = load_iris(data_path).unwrap();

    // encode the target variables
    let mut encoder = OneHotEncoding::new(y_train.clone()).unwrap();
    let y_train_encoded = encoder.transform();

    // create logistic regression model
    let mut log_model = MultiClassLogistic::new(
        &x_train,
        &y_train_encoded,
        softmax,
        0.1
    ).unwrap();

    log_model.sgd(500, true, 5);

    let sample_index = 100;
    let x_test = x_train.batch(5).unwrap();
    let y_test = y_train.batch(5).unwrap();
    let y_pred = log_model.predict(x_test[sample_index].clone());

    println!("Actual: {:?}", y_test[sample_index]);
    println!("Prediction: {:?}", y_pred.values());

    let loss = mse(&y_test[sample_index], &y_pred).unwrap(); 
    println!("LOSS: {:?}", loss);  
}