Merge abb31df into a20bf61

CHANGELOG.md

@@ -1,5 +1,8 @@
 # Changelog
 
+## 0.2.0
+- `DataFrame`: `shuffle` method added
+
 ## 0.1.1
 - `DataFrame`: addSeries method added
 

lib/src/data_frame/data_frame.dart

@@ -169,4 +169,7 @@ abstract class DataFrame implements Serializable {
   /// The method may throw an error if the [DataFrame] contains data that
   /// cannot be converted to numeric representation
   Matrix toMatrix([DType dtype]);
+
+  /// Returns a new [DataFrame] with shuffled rows of this [DataFrame]
+  DataFrame shuffle({int seed});
 }

lib/src/data_frame/data_frame_impl.dart

@@ -3,6 +3,7 @@ import 'package:ml_dataframe/src/data_frame/data_frame.dart';
 import 'package:ml_dataframe/src/data_frame/data_frame_helpers.dart';
 import 'package:ml_dataframe/src/data_frame/data_frame_json_keys.dart';
 import 'package:ml_dataframe/src/data_frame/errors/wrong_series_shape_exception.dart';
+import 'package:ml_dataframe/src/data_frame/helpers/generate_unordered_indices.dart';
 import 'package:ml_dataframe/src/data_frame/series.dart';
 import 'package:ml_dataframe/src/numerical_converter/helpers/from_numerical_converter_json.dart';
 import 'package:ml_dataframe/src/numerical_converter/helpers/numerical_converter_to_json.dart';
@@ -12,6 +13,7 @@ import 'package:ml_linalg/dtype.dart';
 import 'package:ml_linalg/linalg.dart';
 import 'package:ml_linalg/matrix.dart';
 import 'package:quiver/iterables.dart';
+import 'dart:math' as math;
 
 part 'data_frame_impl.g.dart';
 
@@ -135,6 +137,15 @@ class DataFrameImpl with SerializableMixin implements DataFrame {
       dtype: dtype,
     );
 
+  @override
+  DataFrame shuffle({int seed}) {
+    final rowsAsList = rows.toList();
+    final indices = generateUnorderedIndices(shape.first, seed);
+    final shuffledRows = indices.map((index) => rowsAsList[index]);
+
+    return DataFrame(shuffledRows, header: header, headerExists: false);
+  }
+
   DataFrame _sampleFromSeries(Iterable ids) =>
       DataFrame.fromSeries(ids.map((dynamic id) => this[id]));
 

lib/src/data_frame/helpers/generate_unordered_indices.dart

@@ -0,0 +1,23 @@
+import 'dart:math' as math;
+
+import 'package:quiver/iterables.dart';
+
+List<int> generateUnorderedIndices(int length, [int seed]) {
+  if (length <= 0) {
+    return [];
+  }
+
+  final generator = math.Random(seed);
+  final orderedIndices = List.generate(length, (i) => i);
+  final indices = [...orderedIndices]
+      .toList()
+    ..shuffle(generator);
+  final ensureIndicesAreUnordered = () => zip([indices, orderedIndices])
+      .any((pair) => pair.first != pair.last);
+
+  while (!ensureIndicesAreUnordered()) {
+    indices.shuffle(generator);
+  }
+
+  return indices;
+}

pubspec.yaml

@@ -1,6 +1,6 @@
 name: ml_dataframe
 description: Dataframe - a way to store and manipulate data
-version: 0.1.1
+version: 0.2.0
 homepage: https://github.com/gyrdym/ml_dataframe
 
 environment:

test/data_frame/data_frame_test.dart

@@ -331,6 +331,38 @@ void main() {
       });
     });
 
+    group('shuffle', () {
+      test('should return a new dataframe with the same header', () {
+        final dataframe = DataFrame(data);
+        final shuffled = dataframe.shuffle();
+
+        expect(shuffled.header, dataframe.header);
+      });
+
+      test('should return a new dataframe with the same number of rows', () {
+        final dataframe = DataFrame(data);
+        final shuffled = dataframe.shuffle();
+
+        expect(shuffled.rows, hasLength(dataframe.rows.length));
+      });
+
+      test('should return a new dataframe with the different order of rows', () {
+        final dataframe = DataFrame(data);
+        final shuffled = dataframe.shuffle();
+
+        expect(shuffled.rows, isNot(equals(dataframe.rows)));
+      });
+
+      test('should return a new dataframe containing the same rows', () {
+        final dataframe = DataFrame(data);
+        final shuffled = dataframe.shuffle();
+
+        dataframe.rows.forEach((row) {
+          expect(shuffled.rows, contains(equals(row)));
+        });
+      });
+    });
+
     group('serialization', () {
       final json = {
         dataFrameHeaderJsonKey: ['first',  'second', 'third'],