maximal_regression/
maximal_regression.rs

1use automl::settings::*;
2use automl::*;
3
4fn main() {
5    // Totally customize settings
6    let settings = Settings::default_regression()
7        .with_number_of_folds(3)
8        .shuffle_data(true)
9        .verbose(true)
10        .with_final_model(FinalModel::Blending {
11            algorithm: Algorithm::Linear,
12            meta_training_fraction: 0.15,
13            meta_testing_fraction: 0.15,
14        })
15        .skip(Algorithm::RandomForestRegressor)
16        .sorted_by(Metric::RSquared)
17        .with_preprocessing(PreProcessing::AddInteractions)
18        .with_linear_settings(
19            LinearRegressionParameters::default().with_solver(LinearRegressionSolverName::QR),
20        )
21        .with_lasso_settings(
22            LassoParameters::default()
23                .with_alpha(1.0)
24                .with_tol(1e-4)
25                .with_normalize(true)
26                .with_max_iter(1000),
27        )
28        .with_ridge_settings(
29            RidgeRegressionParameters::default()
30                .with_alpha(1.0)
31                .with_normalize(true)
32                .with_solver(RidgeRegressionSolverName::Cholesky),
33        )
34        .with_elastic_net_settings(
35            ElasticNetParameters::default()
36                .with_tol(1e-4)
37                .with_normalize(true)
38                .with_alpha(1.0)
39                .with_max_iter(1000)
40                .with_l1_ratio(0.5),
41        )
42        .with_knn_regressor_settings(
43            KNNRegressorParameters::default()
44                .with_algorithm(KNNAlgorithmName::CoverTree)
45                .with_k(3)
46                .with_distance(Distance::Euclidean)
47                .with_weight(KNNWeightFunction::Uniform),
48        )
49        .with_svr_settings(
50            SVRParameters::default()
51                .with_eps(0.1)
52                .with_tol(1e-3)
53                .with_c(1.0)
54                .with_kernel(Kernel::Linear),
55        )
56        .with_random_forest_regressor_settings(
57            RandomForestRegressorParameters::default()
58                .with_m(1)
59                .with_max_depth(5)
60                .with_min_samples_leaf(1)
61                .with_n_trees(10)
62                .with_min_samples_split(2),
63        )
64        .with_decision_tree_regressor_settings(
65            DecisionTreeRegressorParameters::default()
66                .with_min_samples_split(2)
67                .with_max_depth(15)
68                .with_min_samples_leaf(1),
69        );
70
71    // Save the settings for later use
72    settings.save("examples/maximal_regression_settings.yaml");
73
74    // Load a dataset from smartcore and add it to the regressor along with the customized settings
75    let mut model = SupervisedModel::new(smartcore::dataset::diabetes::load_dataset(), settings);
76
77    // Run a model comparison with all models at default settings
78    model.train();
79
80    // Print the results
81    println!("{}", model);
82
83    // Save teh model for later
84    model.save("examples/maximal_regression_model.aml");
85}
maximal_regression/maximal_regression.rs

maximal_regression/
maximal_regression.rs
