AutoML with Linfa

AutoML is Automated Machine Learning, referring to processes and methods to make machine learning more accesible for a general audience. This crate builds on top of the linfa machine learning framework, and provides some utilities to quickly train and compare models.

Usage

Running the following: rust fn main() { let data = linfa_datasets::diabetes(); let r = automl::regression::compare_models(&data); print!("{}", r); } Will output this: text ┌──────────────┬───────┬─────────┬─────────┬───────────┐ │ Model │ R^2 │ MSE │ MAE │ Exp. Var. │ ╞══════════════╪═══════╪═════════╪═════════╪═══════════╡ │ Linear Model │ 0.519 │ 2.859e3 │ 4.326e1 │ 5.189e-1 │ ├──────────────┼───────┼─────────┼─────────┼───────────┤ │ Elastic Net │ 0.009 │ 5.891e3 │ 6.563e1 │ 8.864e-3 │ ├──────────────┼───────┼─────────┼─────────┼───────────┤ │ LASSO │ 0.359 │ 3.811e3 │ 5.254e1 │ 3.589e-1 │ ├──────────────┼───────┼─────────┼─────────┼───────────┤ │ Ridge │ 0.007 │ 5.904e3 │ 6.571e1 │ 6.537e-3 │ └──────────────┴───────┴─────────┴─────────┴───────────┘ Based on this output, you can then select the best model for the task.

Roadmap

Currently this crate only includes some regression functions, but classification and clustering utilities will be developed in the future. - Regression - [ ] Preprocessing - [ ] Hyperparameter Tuning - [x] ElasticNet - [x] Linear Regression - [ ] PLS Regression - Classification - [ ] Preprocessing - [ ] Hyperparameter Tuning - [ ] Naive Bayes - [ ] SVM - [ ] Decision Trees - [ ] Logistic Regression - Clustering - [ ] Preprocessing - [ ] Hyperparameter Tuning - [ ] K-Means - [ ] DBSCAN - [ ] Approximated DBSCAN - [ ] Gaussian Mixture Model - [ ] Hierarchical