Merge 6ebe3b9 into 4ed244a

CHANGELOG.md

@@ -1,5 +1,8 @@
 # Changelog
 
+## 0.0.10
+- `DataFrame.sampleFromSeries` method's signature changed 
+
 ## 0.0.9
 - `dtype` field added to `DataFrame` interface
 

lib/src/data_frame/data_frame.dart

@@ -92,7 +92,7 @@ abstract class DataFrame {
 
   Iterable<Iterable<dynamic>> get rows;
 
-  /// Returns series (columns) collection of the [DataFrame].
+  /// Returns a lazy series (columns) collection of the [DataFrame].
   ///
   /// [Series] is roughly a column and its header (name)
   Iterable<Series> get series;
@@ -103,14 +103,15 @@ abstract class DataFrame {
   /// series)
   Series operator [](Object key);
 
-  /// Returns a collection of dataframes, sampled from series, that are
-  /// obtained from provided [indices] parts or [names] parts.
+  /// Returns a dataframe, sampled from series, that are obtained from provided
+  /// series [indices] or series [names].
   ///
-  /// Series id (index or name) may repeat in different parts but not in the
-  /// same part
-  Iterable<DataFrame> sampleFromSeries({
-    Iterable<Iterable<int>> indices,
-    Iterable<Iterable<String>> names,
+  /// If [indices] are specified, [names] parameter will be ignored.
+  ///
+  /// Series indices or series names may be repeating.
+  DataFrame sampleFromSeries({
+    Iterable<int> indices,
+    Iterable<String> names,
   });
 
   /// Returns a new [DataFrame] without specified series (columns)

lib/src/data_frame/data_frame_impl.dart

@@ -56,13 +56,26 @@ class DataFrameImpl implements DataFrame {
   }
 
   @override
-  Iterable<DataFrame> sampleFromSeries({
-    Iterable<Iterable<int>> indices = const [],
-    Iterable<Iterable<String>> names = const [],
+  DataFrame sampleFromSeries({
+    Iterable<int> indices = const [],
+    Iterable<String> names = const [],
   }) {
     if (indices.isNotEmpty) {
+      final maxIdx = series.length - 1;
+      final outRangedIndices = indices.where((idx) => idx < 0 || idx > maxIdx);
+      if (outRangedIndices.isNotEmpty) {
+        throw RangeError('Some of provided indices are out of range: '
+            '$outRangedIndices, while the valid range is 0..$maxIdx (both '
+            'inclusive)');
+      }
       return _sampleFromSeries(indices);
     }
+    final absentNames = Set<String>
+        .from(names)
+        .difference(Set.from(header));
+    if (absentNames.isNotEmpty) {
+      throw Exception('Columns with the names $absentNames do not exist');
+    };
     return _sampleFromSeries(names);
   }
 
@@ -88,11 +101,11 @@ class DataFrameImpl implements DataFrame {
         dtype: dtype,
       );
 
-  Iterable<DataFrame> _sampleFromSeries(Iterable<Iterable> allIds) =>
-      allIds.map((ids) {
-        final uniqueIds = Set<dynamic>.from(ids);
-        return DataFrame.fromSeries(uniqueIds.map((dynamic id) => this[id]));
-      });
+  DataFrame _sampleFromSeries(Iterable ids) =>
+      DataFrame.fromSeries(
+        ids.map((dynamic id) => this[id]),
+        dtype: dtype,
+      );
 
   DataFrame _dropByIndices(Iterable<int> indices, Iterable<Series> series) {
     final uniqueIndices = Set<int>.from(indices);

lib/src/data_frame/series.dart

@@ -5,9 +5,17 @@ class Series {
       ? Set<dynamic>.from(data)
       : const <dynamic>[];
 
+  /// A name of the [Series]
   final String name;
+
+  /// Return a lazy iterable of data, containing in the [Series]
   final Iterable data;
+
+  /// Returns a lazy iterable of the [data]'s unique values if the [Series]
+  /// marked as [isDiscrete]. If [isDiscrete] is `false`, an empty list will be
+  /// returned
   final Iterable discreteValues;
 
+  /// Shows if [Series] contains just discrete values instead of continuous ones
   bool get isDiscrete => discreteValues.isNotEmpty;
 }

pubspec.yaml

@@ -1,6 +1,6 @@
 name: ml_dataframe
 description: Dataframe - a way to store and manipulate data
-version: 0.0.9
+version: 0.0.10
 author: Ilia Gyrdymov <ilgyrd@gmail.com>
 homepage: https://github.com/gyrdym/ml_dataframe
 

test/data_frame/data_frame_test.dart

@@ -193,122 +193,108 @@ void main() {
         ];
 
         final dataFrame = DataFrame(data, headerExists: false);
-        final dataFrames = dataFrame.sampleFromSeries(indices: [
-          [0, 1],
-          [2, 3],
-          [4],
-        ]).toList();
-
-        expect(dataFrames, hasLength(3));
-
-        expect(dataFrames[0].rows, equals([
-          [  '1',       2 ],
-          [   10,      12 ],
-          [  -10,     202 ],
-        ]));
+        final sampled = dataFrame.sampleFromSeries(indices: [0, 1, 2, 3, 4]);
 
-        expect(dataFrames[1].rows, equals([
-          [    3,        0  ],
-          [  323,      1.5  ],
-          [ 1000,     '1.5' ],
+        expect(sampled.rows, equals([
+          [  '1',       2,        3,        0,         32 ],
+          [   10,      12,      323,      1.5,       1132 ],
+          [  -10,     202,     1000,     '1.5',     0.005 ],
         ]));
+      });
+
+      test('should support repeating indices', () {
+        final data = [
+          [  '1',       2,        3,        0,         32 ],
+          [   10,      12,      323,      1.5,       1132 ],
+          [  -10,     202,     1000,     '1.5',     0.005 ],
+        ];
 
-        expect(dataFrames[2].rows, equals([
-          [    32 ],
-          [  1132 ],
-          [ 0.005 ],
+        final dataFrame = DataFrame(data, headerExists: false);
+        final sampled = dataFrame.sampleFromSeries(indices: [0, 1, 0, 1]);
+
+        expect(sampled.rows, equals([
+          [  '1',   2,  '1',  2 ],
+          [   10,  12,  10,  12 ],
+          [  -10, 202, -10, 202 ],
         ]));
       });
 
-      test('should support repeating indices in different parts', () {
+      test('should sample dataframe by series names', () {
         final data = [
           [  '1',       2,        3,        0,         32 ],
           [   10,      12,      323,      1.5,       1132 ],
           [  -10,     202,     1000,     '1.5',     0.005 ],
         ];
 
         final dataFrame = DataFrame(data, headerExists: false);
-        final dataFrames = dataFrame.sampleFromSeries(indices: [
-          [0, 1],
-          [0, 1],
-        ]).toList();
+        final sampled = dataFrame.sampleFromSeries(
+            names: ['col_0', 'col_1', 'col_2', 'col_3', 'col_4']);
 
-        expect(dataFrames, hasLength(2));
-
-        expect(dataFrames[0].rows, equals([
-          [  '1',       2 ],
-          [   10,      12 ],
-          [  -10,     202 ],
+        expect(sampled.rows, equals([
+          [  '1',       2,        3,        0,         32 ],
+          [   10,      12,      323,      1.5,       1132 ],
+          [  -10,     202,     1000,     '1.5',     0.005 ],
         ]));
+      });
 
-        expect(dataFrames[1].rows, equals([
-          [  '1',       2 ],
-          [   10,      12 ],
-          [  -10,     202 ],
+      test('should ignore names parameter if indices parameter is '
+          'provided', () {
+        final data = [
+          [  '1',       2,        3,        0,         32 ],
+          [   10,      12,      323,      1.5,       1132 ],
+          [  -10,     202,     1000,     '1.5',     0.005 ],
+        ];
+
+        final dataFrame = DataFrame(data, headerExists: false);
+        final sampled = dataFrame.sampleFromSeries(
+            indices: [1],
+            names: ['col_0', 'col_1', 'col_2', 'col_3', 'col_4']);
+
+        expect(sampled.rows, equals([
+          [   2, ],
+          [  12, ],
+          [ 202, ],
         ]));
       });
 
-      test('should skip repeating indices in the same part', () {
+      test('should throw an error if outranged indices are provided', () {
         final data = [
           [  '1',       2,        3,        0,         32 ],
           [   10,      12,      323,      1.5,       1132 ],
           [  -10,     202,     1000,     '1.5',     0.005 ],
         ];
 
         final dataFrame = DataFrame(data, headerExists: false);
-        final dataFrames = dataFrame.sampleFromSeries(indices: [
-          [0, 0],
-          [2, 2],
-        ]).toList();
+        final actual = () => dataFrame.sampleFromSeries(indices: [1, 40]);
 
-        expect(dataFrames, hasLength(2));
+        expect(actual, throwsRangeError);
+      });
 
-        expect(dataFrames[0].rows, equals([
-          [   '1' ],
-          [   10  ],
-          [  -10  ],
-        ]));
+      test('should throw an error if outranged negative indices are provided', () {
+        final data = [
+          [  '1',       2,        3,        0,         32 ],
+          [   10,      12,      323,      1.5,       1132 ],
+          [  -10,     202,     1000,     '1.5',     0.005 ],
+        ];
 
-        expect(dataFrames[1].rows, equals([
-          [     3 ],
-          [   323 ],
-          [  1000 ],
-        ]));
+        final dataFrame = DataFrame(data, headerExists: false);
+        final actual = () => dataFrame.sampleFromSeries(indices: [2, -1]);
+
+        expect(actual, throwsRangeError);
       });
 
-      test('should sample dataframe by series names', () {
+      test('should throw an error if names of non existent columns are '
+          'provided', () {
         final data = [
           [  '1',       2,        3,        0,         32 ],
           [   10,      12,      323,      1.5,       1132 ],
           [  -10,     202,     1000,     '1.5',     0.005 ],
         ];
 
         final dataFrame = DataFrame(data, headerExists: false);
-        final dataFrames = dataFrame.sampleFromSeries(names: [
-          ['col_0', 'col_1'],
-          ['col_2', 'col_3'],
-          ['col_4'],
-        ]).toList();
-
-        expect(dataFrames, hasLength(3));
-
-        expect(dataFrames[0].rows, equals([
-          [  '1',       2 ],
-          [   10,      12 ],
-          [  -10,     202 ],
-        ]));
+        final actual = () => dataFrame.sampleFromSeries(names: ['col_0', 'col_100']);
 
-        expect(dataFrames[1].rows, equals([
-          [    3,        0  ],
-          [  323,      1.5  ],
-          [ 1000,     '1.5' ],
-        ]));
-
-        expect(dataFrames[2].rows, equals([
-          [    32 ],
-          [  1132 ],
-          [ 0.005 ],
-        ]));
+        expect(actual, throwsException);
       });
     });
   });