Dataset improvements (#74)

* Renamed Dataset to DatasetBase in all subcrates. Added new types Dataset and DatasetView

* impl all method except k-fold, update dataset crate

* Dropped vec implementation. Adjusted svm accordingly

* k-folding with axis_chunks_iter

* renamed Dataset to DatasetBase in naive bayes

* swapping for k-fold, dropped vec impl

datasets/src/lib.rs

@@ -21,7 +21,7 @@ fn array_from_buf(buf: &[u8]) -> Array2<f64> {
 #[cfg(feature = "iris")]
 /// Read in the iris-flower dataset from dataset path
 /// The `.csv` data is two dimensional: Axis(0) denotes y-axis (rows), Axis(1) denotes x-axis (columns)
-pub fn iris() -> Dataset<Array2<f64>, Vec<usize>> {
+pub fn iris() -> Dataset<f64, usize> {
     let data = include_bytes!("../data/iris.csv.gz");
     let array = array_from_buf(&data[..]);
 
@@ -34,7 +34,7 @@ pub fn iris() -> Dataset<Array2<f64>, Vec<usize>> {
 }
 
 #[cfg(feature = "diabetes")]
-pub fn diabetes() -> Dataset<Array2<f64>, Array1<f64>> {
+pub fn diabetes() -> Dataset<f64, f64> {
     let data = include_bytes!("../data/diabetes_data.csv.gz");
     let data = array_from_buf(&data[..]);
 
@@ -45,7 +45,7 @@ pub fn diabetes() -> Dataset<Array2<f64>, Array1<f64>> {
 }
 
 #[cfg(feature = "winequality")]
-pub fn winequality() -> Dataset<Array2<f64>, Vec<usize>> {
+pub fn winequality() -> Dataset<f64, usize> {
     let data = include_bytes!("../data/winequality-red.csv.gz");
     let array = array_from_buf(&data[..]);
 

linfa-bayes/src/gaussian_nb.rs

@@ -9,7 +9,7 @@ use ndarray_stats::QuantileExt;
 use std::collections::HashMap;
 
 use crate::error::Result;
-use linfa::dataset::{Dataset, Labels};
+use linfa::dataset::{DatasetBase, Labels};
 use linfa::traits::{Fit, IncrementalFit, Predict};
 use linfa::Float;
 
@@ -55,7 +55,7 @@ where
     ///
     /// ```no_run
     /// # use ndarray::array;
-    /// # use linfa::Dataset;
+    /// # use linfa::DatasetBase;
     /// # use linfa_bayes::GaussianNbParams;
     /// # use linfa::traits::{Fit, Predict};
     /// # use std::error::Error;
@@ -70,15 +70,15 @@ where
     /// ];
     /// let y = vec![1, 1, 1, 2, 2, 2];
     ///
-    /// let data = Dataset::new(x.view(), &y);
+    /// let data = DatasetBase::new(x.view(), &y);
     /// let model = GaussianNbParams::params().fit(&data)?;
     /// let pred = model.predict(x.view());
     ///
     /// assert_eq!(pred.to_vec(), y);
     /// # Ok(())
     /// # }
     /// ```
-    fn fit(&self, dataset: &'a Dataset<ArrayView2<A>, L>) -> Self::Object {
+    fn fit(&self, dataset: &'a DatasetBase<ArrayView2<A>, L>) -> Self::Object {
         // We extract the unique classes in sorted order
         let mut unique_classes = dataset.targets.labels();
         unique_classes.sort_unstable();
@@ -106,7 +106,7 @@ where
     ///
     /// ```no_run
     /// # use ndarray::{array, Axis};
-    /// # use linfa::Dataset;
+    /// # use linfa::DatasetBase;
     /// # use linfa_bayes::GaussianNbParams;
     /// # use linfa::traits::{Predict, IncrementalFit};
     /// # use std::error::Error;
@@ -128,7 +128,7 @@ where
     ///     .axis_chunks_iter(Axis(0), 2)
     ///     .zip(y.axis_chunks_iter(Axis(0), 2))
     /// {
-    ///     model = clf.fit_with(model, &Dataset::new(x, y))?;
+    ///     model = clf.fit_with(model, &DatasetBase::new(x, y))?;
     /// }
     ///
     /// let pred = model.as_ref().unwrap().predict(x.view());
@@ -140,7 +140,7 @@ where
     fn fit_with(
         &self,
         model_in: Self::ObjectIn,
-        dataset: &Dataset<ArrayView2<A>, L>,
+        dataset: &DatasetBase<ArrayView2<A>, L>,
     ) -> Self::ObjectOut {
         let x = dataset.records();
         let y = dataset.targets();
@@ -358,7 +358,7 @@ impl<A: Float> GaussianNb<A> {
 mod tests {
     use super::*;
     use approx::assert_abs_diff_eq;
-    use linfa::Dataset;
+    use linfa::DatasetBase;
     use ndarray::array;
 
     #[test]
@@ -374,7 +374,7 @@ mod tests {
         let y = array![1, 1, 1, 2, 2, 2];
 
         let clf = GaussianNbParams::params();
-        let data = Dataset::new(x.view(), y.view());
+        let data = DatasetBase::new(x.view(), y.view());
         let fitted_clf = clf.fit(&data).unwrap();
         let pred = fitted_clf.predict(x.view());
         assert_eq!(pred, y);
@@ -424,7 +424,7 @@ mod tests {
         let model = x
             .axis_chunks_iter(Axis(0), 2)
             .zip(y.axis_chunks_iter(Axis(0), 2))
-            .map(|(a, b)| Dataset::new(a, b))
+            .map(|(a, b)| DatasetBase::new(a, b))
             .fold(None, |current, d| clf.fit_with(current, &d).unwrap())
             .unwrap();
 

linfa-clustering/benches/gaussian_mixture.rs

@@ -3,7 +3,7 @@ use criterion::{
     PlotConfiguration,
 };
 use linfa::traits::Fit;
-use linfa::Dataset;
+use linfa::DatasetBase;
 use linfa_clustering::{generate_blobs, GaussianMixtureModel};
 use ndarray::Array2;
 use ndarray_rand::rand::SeedableRng;
@@ -22,7 +22,8 @@ fn gaussian_mixture_bench(c: &mut Criterion) {
             let n_features = 3;
             let centroids =
                 Array2::random_using((n_clusters, n_features), Uniform::new(-30., 30.), &mut rng);
-            let dataset = Dataset::from(generate_blobs(cluster_size, &centroids, &mut rng));
+            let dataset: DatasetBase<_, _> =
+                (generate_blobs(cluster_size, &centroids, &mut rng), ()).into();
             bencher.iter(|| {
                 black_box(
                     GaussianMixtureModel::params(n_clusters)

linfa-clustering/benches/k_means.rs

@@ -3,7 +3,7 @@ use criterion::{
     PlotConfiguration,
 };
 use linfa::traits::Fit;
-use linfa::Dataset;
+use linfa::DatasetBase;
 use linfa_clustering::{generate_blobs, KMeans};
 use ndarray::Array2;
 use ndarray_rand::rand::SeedableRng;
@@ -22,7 +22,7 @@ fn k_means_bench(c: &mut Criterion) {
             let n_features = 3;
             let centroids =
                 Array2::random_using((n_clusters, n_features), Uniform::new(-30., 30.), &mut rng);
-            let dataset = Dataset::from(generate_blobs(cluster_size, &centroids, &mut rng));
+            let dataset = DatasetBase::from(generate_blobs(cluster_size, &centroids, &mut rng));
             bencher.iter(|| {
                 black_box(
                     KMeans::params_with_rng(n_clusters, rng.clone())

linfa-clustering/examples/kmeans.rs

@@ -1,5 +1,5 @@
 use linfa::traits::{Fit, Predict};
-use linfa::Dataset;
+use linfa::DatasetBase;
 use linfa_clustering::{generate_blobs, KMeans};
 use ndarray::{array, Axis};
 use ndarray_npy::write_npy;
@@ -15,7 +15,7 @@ fn main() {
     // For each our expected centroids, generate `n` data points around it (a "blob")
     let expected_centroids = array![[10., 10.], [1., 12.], [20., 30.], [-20., 30.],];
     let n = 10000;
-    let dataset = Dataset::from(generate_blobs(n, &expected_centroids, &mut rng));
+    let dataset = DatasetBase::from(generate_blobs(n, &expected_centroids, &mut rng));
 
     // Configure our training algorithm
     let n_clusters = expected_centroids.len_of(Axis(0));
@@ -27,7 +27,7 @@ fn main() {
 
     // Assign each point to a cluster using the set of centroids found using `fit`
     let dataset = model.predict(dataset);
-    let Dataset {
+    let DatasetBase {
         records, targets, ..
     } = dataset;
 

linfa-clustering/src/appx_dbscan/algorithm.rs

@@ -2,7 +2,7 @@ use crate::appx_dbscan::clustering::AppxDbscanLabeler;
 use crate::appx_dbscan::hyperparameters::{AppxDbscanHyperParams, AppxDbscanHyperParamsBuilder};
 use linfa::dataset::Targets;
 use linfa::traits::Transformer;
-use linfa::{Dataset, Float};
+use linfa::{DatasetBase, Float};
 use ndarray::{Array1, ArrayBase, Data, Ix2};
 #[cfg(feature = "serde")]
 use serde_crate::{Deserialize, Serialize};
@@ -107,13 +107,15 @@ impl<F: Float, D: Data<Elem = F>> Transformer<&ArrayBase<D, Ix2>, Array1<Option<
 }
 
 impl<F: Float, D: Data<Elem = F>, T: Targets>
-    Transformer<Dataset<ArrayBase<D, Ix2>, T>, Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>>>
-    for AppxDbscanHyperParams<F>
+    Transformer<
+        DatasetBase<ArrayBase<D, Ix2>, T>,
+        DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>>,
+    > for AppxDbscanHyperParams<F>
 {
     fn transform(
         &self,
-        dataset: Dataset<ArrayBase<D, Ix2>, T>,
-    ) -> Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
+        dataset: DatasetBase<ArrayBase<D, Ix2>, T>,
+    ) -> DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
         let predicted = self.transform(dataset.records());
         dataset.with_targets(predicted)
     }
@@ -128,13 +130,15 @@ impl<F: Float, D: Data<Elem = F>> Transformer<&ArrayBase<D, Ix2>, Array1<Option<
 }
 
 impl<F: Float, D: Data<Elem = F>, T: Targets>
-    Transformer<Dataset<ArrayBase<D, Ix2>, T>, Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>>>
-    for AppxDbscanHyperParamsBuilder<F>
+    Transformer<
+        DatasetBase<ArrayBase<D, Ix2>, T>,
+        DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>>,
+    > for AppxDbscanHyperParamsBuilder<F>
 {
     fn transform(
         &self,
-        dataset: Dataset<ArrayBase<D, Ix2>, T>,
-    ) -> Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
+        dataset: DatasetBase<ArrayBase<D, Ix2>, T>,
+    ) -> DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
         self.build().transform(dataset)
     }
 }

linfa-clustering/src/dbscan/algorithm.rs

@@ -4,7 +4,7 @@ use ndarray_stats::DeviationExt;
 
 use linfa::dataset::Targets;
 use linfa::traits::Transformer;
-use linfa::{Dataset, Float};
+use linfa::{DatasetBase, Float};
 
 #[derive(Clone, Debug, PartialEq)]
 /// DBSCAN (Density-based Spatial Clustering of Applications with Noise)
@@ -115,13 +115,15 @@ impl<F: Float, D: Data<Elem = F>> Transformer<&ArrayBase<D, Ix2>, Array1<Option<
 }
 
 impl<F: Float, D: Data<Elem = F>, T: Targets>
-    Transformer<Dataset<ArrayBase<D, Ix2>, T>, Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>>>
-    for DbscanHyperParams<F>
+    Transformer<
+        DatasetBase<ArrayBase<D, Ix2>, T>,
+        DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>>,
+    > for DbscanHyperParams<F>
 {
     fn transform(
         &self,
-        dataset: Dataset<ArrayBase<D, Ix2>, T>,
-    ) -> Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
+        dataset: DatasetBase<ArrayBase<D, Ix2>, T>,
+    ) -> DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
         let predicted = self.transform(dataset.records());
         dataset.with_targets(predicted)
     }
@@ -136,13 +138,15 @@ impl<F: Float, D: Data<Elem = F>> Transformer<&ArrayBase<D, Ix2>, Array1<Option<
 }
 
 impl<F: Float, D: Data<Elem = F>, T: Targets>
-    Transformer<Dataset<ArrayBase<D, Ix2>, T>, Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>>>
-    for DbscanHyperParamsBuilder<F>
+    Transformer<
+        DatasetBase<ArrayBase<D, Ix2>, T>,
+        DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>>,
+    > for DbscanHyperParamsBuilder<F>
 {
     fn transform(
         &self,
-        dataset: Dataset<ArrayBase<D, Ix2>, T>,
-    ) -> Dataset<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
+        dataset: DatasetBase<ArrayBase<D, Ix2>, T>,
+    ) -> DatasetBase<ArrayBase<D, Ix2>, Array1<Option<usize>>> {
         self.build().transform(dataset)
     }
 }