Crate rustrees

Rustrees is a library for building decision trees and random forests.

The goal is to provide a fast implementation of decision trees in rust, with a python API.

Example usage:

use rustrees::{DecisionTree, Dataset, r2};
 
let dataset = Dataset::read_csv("datasets/titanic_train.csv", ",");
 
let dt = DecisionTree::train_reg(
   &dataset, 
   Some(5),        // max_depth
   Some(1),  // min_samples_leaf        
   None,     // max_features (None = all features)
   Some(42), // random_state
);
 
let pred = dt.predict(&dataset);
 
println!("r2 score: {}", r2(&dataset.target_vector, &pred));

Structs

Dataset

Dataset represents the data used to train the model.

DecisionTree

Represents the decision tree model. Each node represents a split on a feature.

RandomForest

Represents the Random forest model. It is basically a collection of decision trees.

TrainOptions

Possible options for training the model.

Tree

An arena-based tree implementation. Each node is stored in a vector and the children are accessed by index.

Functions

accuracy

computes the accuracy of a binary classification. Used for testing.

r2

computes the mean squared error between two vectors used for testing regression case.