Introduce checked parameters to all algorithms (rust-ml#158)

* Introduce param guards to linfa

* Add param guards to linfa-bayes + documentation

* Add param guards to linfa-elasticnet + docs

* Add remaining changes

* Reset logistic regression

* Add ParamGuard to HierarchicalCluster

* Fix PLS errors

* Add ParamGuard to logistic regression

* Fixed logistic regression tests

* Fix Clippy warnings

* Tweak logistic regression

* Fix elasticnet doctests

* Fix doctests in linfa-bayes

* Fix Gaussian mixture example

* Fix tsne examples

* Fix SVM examples

* Simplify reduction transform impl return type

* Resolve transform double Result issue using marker traits

Co-authored-by: Lorenz Schmidt <[email protected]>
Co-authored-by: YuhanLiin <[email protected]>

Author: 3 people

.github/workflows/checking.yml

@@ -11,7 +11,7 @@ jobs:
       fail-fast: false
       matrix:
         toolchain:
-          - 1.51.0
+          - 1.54.0
           - stable
         experimental: [false]
         include:
@@ -37,7 +37,7 @@ jobs:
           args: --workspace
 
       - name: Run cargo check (with serde)
-        if: ${{ matrix.toolchain != '1.51.0' }} # The following args don't work on older versions of Cargo
+        if: ${{ matrix.toolchain != '1.54.0' }} # The following args don't work on older versions of Cargo
         uses: actions-rs/cargo@v1
         with:
           command: check

.github/workflows/codequality.yml

@@ -11,7 +11,7 @@ jobs:
     strategy:
       matrix:
         toolchain:
-          - 1.51.0
+          - 1.54.0
           - stable
 
     steps:

.github/workflows/testing.yml

@@ -14,7 +14,7 @@ jobs:
       fail-fast: false
       matrix:
         toolchain:
-          - 1.51.0
+          - 1.54.0
           - stable
 
     steps:

Cargo.toml

@@ -74,21 +74,7 @@ linfa-datasets = { path = "datasets", features = ["winequality", "iris", "diabet
 
 [workspace]
 members = [
-    "algorithms/linfa-clustering",
-    "algorithms/linfa-reduction",
-    "algorithms/linfa-kernel",
-    "algorithms/linfa-linear",
-    "algorithms/linfa-logistic",
-    "algorithms/linfa-trees",
-    "algorithms/linfa-svm",
-    "algorithms/linfa-hierarchical",
-    "algorithms/linfa-ica",
-    "algorithms/linfa-bayes",
-    "algorithms/linfa-elasticnet",
-    "algorithms/linfa-pls",
-    "algorithms/linfa-tsne",
-    "algorithms/linfa-preprocessing",
-    "algorithms/linfa-nn",
+    "algorithms/*",
     "datasets",
 ]
 

algorithms/linfa-bayes/README.md

@@ -1,6 +1,6 @@
 # Naive Bayes
 
-`linfa-bayes` aims to provide pure Rust implementations of Naive Bayes algorithms. 
+`linfa-bayes` provides pure Rust implementations of Naive Bayes algorithms for the Linfa toolkit.
 
 ## The Big Picture
 
@@ -14,14 +14,43 @@
 
 ## Examples
 
-There is an usage example in the `examples/` directory. To run, use:
+You can find an example in the `examples/` directory. To run, use:
 
 ```bash
-$ cargo run --example winequality 
+$ cargo run --example winequality --release
 ```
 
-## License
-Dual-licensed to be compatible with the Rust project.
-
-Licensed under the Apache License, Version 2.0 <http://www.apache.org/licenses/LICENSE-2.0> or the MIT license <http://opensource.org/licenses/MIT>, at your option. This file may not be copied, modified, or distributed except according to those terms.
-
+<details>
+<summary style="cursor: pointer; display:list-item;">
+Show source code
+</summary>
+
+```rust, no_run
+use linfa::metrics::ToConfusionMatrix;
+use linfa::traits::{Fit, Predict};
+use linfa_bayes::{GaussianNb, Result};
+
+// Read in the dataset and convert targets to binary data
+let (train, valid) = linfa_datasets::winequality()
+    .map_targets(|x| if *x > 6 { "good" } else { "bad" })
+    .split_with_ratio(0.9);
+
+// Train the model
+let model = GaussianNb::params().fit(&train)?;
+
+// Predict the validation dataset
+let pred = model.predict(&valid);
+
+// Construct confusion matrix
+let cm = pred.confusion_matrix(&valid)?;
+
+// classes    | bad        | good      
+// bad        | 130        | 12        
+// good       | 7          | 10    
+//
+// accuracy 0.8805031, MCC 0.45080978
+println!("{:?}", cm);
+println!("accuracy {}, MCC {}", cm.accuracy(), cm.mcc());
+# Result::Ok(())
+```
+</details>

algorithms/linfa-bayes/examples/winequality.rs

@@ -1,25 +1,25 @@
 use linfa::metrics::ToConfusionMatrix;
 use linfa::traits::{Fit, Predict};
-use linfa_bayes::{GaussianNbParams, Result};
+use linfa_bayes::{GaussianNb, Result};
 
 fn main() -> Result<()> {
-    // Read in the dataset and convert continuous target into categorical
+    // Read in the dataset and convert targets to binary data
     let (train, valid) = linfa_datasets::winequality()
-        .map_targets(|x| if *x > 6 { 1 } else { 0 })
+        .map_targets(|x| if *x > 6 { "good" } else { "bad" })
         .split_with_ratio(0.9);
 
     // Train the model
-    let model = GaussianNbParams::params().fit(&train.view())?;
+    let model = GaussianNb::params().fit(&train)?;
 
     // Predict the validation dataset
     let pred = model.predict(&valid);
 
     // Construct confusion matrix
     let cm = pred.confusion_matrix(&valid)?;
 
-    // classes    | 1          | 0
-    // 1          | 10         | 12
-    // 0          | 7          | 130
+    // classes    | bad        | good
+    // bad        | 130        | 12
+    // good       | 7          | 10
     //
     // accuracy 0.8805031, MCC 0.45080978
     println!("{:?}", cm);

algorithms/linfa-bayes/src/error.rs

@@ -1,14 +1,18 @@
 use ndarray_stats::errors::MinMaxError;
 use thiserror::Error;
 
-pub type Result<T> = std::result::Result<T, BayesError>;
+/// Simplified `Result` using [`NaiveBayesError`](crate::NaiveBayesError) as error type
+pub type Result<T> = std::result::Result<T, NaiveBayesError>;
 
-/// An error when using a GaussianNB classifier
+/// Error variants from hyper-parameter construction or model estimation
 #[derive(Error, Debug)]
-pub enum BayesError {
+pub enum NaiveBayesError {
     /// Error when performing Max operation on data
     #[error("invalid statistical operation {0}")]
     Stats(#[from] MinMaxError),
+    /// Invalid smoothing parameter
+    #[error("invalid smoothing parameter {0}")]
+    InvalidSmoothing(f64),
     #[error(transparent)]
     BaseCrate(#[from] linfa::Error),
 }