DataOperationsCatalog.CrossValidationSplit Method

Definition

Namespace:

Microsoft.ML

Assembly:

Microsoft.ML.Data.dll

Package:

Microsoft.ML v3.0.1

Split the dataset into cross-validation folds of train set and test set. Respects the samplingKeyColumnName if provided.

public System.Collections.Generic.IReadOnlyList<Microsoft.ML.DataOperationsCatalog.TrainTestData> CrossValidationSplit (Microsoft.ML.IDataView data, int numberOfFolds = 5, string samplingKeyColumnName = default, int? seed = default);

member this.CrossValidationSplit : Microsoft.ML.IDataView * int * string * Nullable<int> -> System.Collections.Generic.IReadOnlyList<Microsoft.ML.DataOperationsCatalog.TrainTestData>

Public Function CrossValidationSplit (data As IDataView, Optional numberOfFolds As Integer = 5, Optional samplingKeyColumnName As String = Nothing, Optional seed As Nullable(Of Integer) = Nothing) As IReadOnlyList(Of DataOperationsCatalog.TrainTestData)

Parameters

data

IDataView

The dataset to split.

numberOfFolds

Int32

Number of cross-validation folds.

samplingKeyColumnName

String

Name of a column to use for grouping rows. If two examples share the same value of the samplingKeyColumnName, they are guaranteed to appear in the same subset (train or test). This can be used to ensure no label leakage from the train to the test set. Note that when performing a Ranking Experiment, the samplingKeyColumnName must be the GroupId column. If null no row grouping will be performed.

seed

Nullable<Int32>

Seed for the random number generator used to select rows for cross-validation folds.

Returns

IReadOnlyList<TrainTestData>

Examples

using System;
using System.Collections.Generic;
using Microsoft.ML;

namespace Samples.Dynamic
{
    /// <summary>
    /// Sample class showing how to use CrossValidationSplit.
    /// </summary>
    public static class CrossValidationSplit
    {
        public static void Example()
        {
            // Creating the ML.Net IHostEnvironment object, needed for the pipeline.
            var mlContext = new MLContext();

            // Generate some data points.
            var examples = GenerateRandomDataPoints(10);

            // Convert the examples list to an IDataView object, which is consumable
            // by ML.NET API.
            var dataview = mlContext.Data.LoadFromEnumerable(examples);

            // Cross validation splits your data randomly into set of "folds", and
            // creates groups of Train and Test sets, where for each group, one fold
            // is the Test and the rest of the folds the Train. So below, we specify
            // Group column as the column containing the sampling keys. If we pass
            // that column to cross validation it would be used to break data into
            // certain chunks.
            var folds = mlContext.Data
                .CrossValidationSplit(dataview, numberOfFolds: 3,
                samplingKeyColumnName: "Group");

            var trainSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[0].TrainSet,
                reuseRowObject: false);

            var testSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[0].TestSet,
                reuseRowObject: false);

            PrintPreviewRows(trainSet, testSet);

            // The data in the Train split.
            // [Group, 1], [Features, 0.8173254]
            // [Group, 2], [Features, 0.7680227]
            // [Group, 1], [Features, 0.2060332]
            // [Group, 2], [Features, 0.5588848]
            // [Group, 1], [Features, 0.4421779]
            // [Group, 2], [Features, 0.9775497]
            // 
            // The data in the Test split.
            // [Group, 0], [Features, 0.7262433]
            // [Group, 0], [Features, 0.5581612]
            // [Group, 0], [Features, 0.9060271]
            // [Group, 0], [Features, 0.2737045]

            trainSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[1].TrainSet,
                reuseRowObject: false);

            testSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[1].TestSet,
                reuseRowObject: false);

            PrintPreviewRows(trainSet, testSet);
            // The data in the Train split.
            // [Group, 0], [Features, 0.7262433]
            // [Group, 2], [Features, 0.7680227]
            // [Group, 0], [Features, 0.5581612]
            // [Group, 2], [Features, 0.5588848]
            // [Group, 0], [Features, 0.9060271]
            // [Group, 2], [Features, 0.9775497]
            // [Group, 0], [Features, 0.2737045]
            // 
            // The data in the Test split.
            // [Group, 1], [Features, 0.8173254]
            // [Group, 1], [Features, 0.2060332]
            // [Group, 1], [Features, 0.4421779]

            trainSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[2].TrainSet,
                reuseRowObject: false);

            testSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[2].TestSet,
                reuseRowObject: false);

            PrintPreviewRows(trainSet, testSet);
            // The data in the Train split.
            // [Group, 0], [Features, 0.7262433]
            // [Group, 1], [Features, 0.8173254]
            // [Group, 0], [Features, 0.5581612]
            // [Group, 1], [Features, 0.2060332]
            // [Group, 0], [Features, 0.9060271]
            // [Group, 1], [Features, 0.4421779]
            // [Group, 0], [Features, 0.2737045]
            // 
            // The data in the Test split.
            // [Group, 2], [Features, 0.7680227]
            // [Group, 2], [Features, 0.5588848]
            // [Group, 2], [Features, 0.9775497]

            // Example of a split without specifying a sampling key column.
            folds = mlContext.Data.CrossValidationSplit(dataview, numberOfFolds: 3);
            trainSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[0].TrainSet,
                reuseRowObject: false);

            testSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[0].TestSet,
                reuseRowObject: false);

            PrintPreviewRows(trainSet, testSet);
            // The data in the Train split.
            // [Group, 0], [Features, 0.7262433]
            // [Group, 1], [Features, 0.8173254]
            // [Group, 2], [Features, 0.7680227]
            // [Group, 0], [Features, 0.5581612]
            // [Group, 1], [Features, 0.2060332]
            // [Group, 1], [Features, 0.4421779]
            // [Group, 2], [Features, 0.9775497]
            // [Group, 0], [Features, 0.2737045]
            // 
            // The data in the Test split.
            // [Group, 2], [Features, 0.5588848]
            // [Group, 0], [Features, 0.9060271]

            trainSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[1].TrainSet,
                reuseRowObject: false);

            testSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[1].TestSet,
                reuseRowObject: false);

            PrintPreviewRows(trainSet, testSet);
            // The data in the Train split.
            // [Group, 2], [Features, 0.7680227]
            // [Group, 0], [Features, 0.5581612]
            // [Group, 1], [Features, 0.2060332]
            // [Group, 2], [Features, 0.5588848]
            // [Group, 0], [Features, 0.9060271]
            // [Group, 1], [Features, 0.4421779]
            // 
            // The data in the Test split.
            // [Group, 0], [Features, 0.7262433]
            // [Group, 1], [Features, 0.8173254]
            // [Group, 2], [Features, 0.9775497]
            // [Group, 0], [Features, 0.2737045]

            trainSet = mlContext.Data
                .CreateEnumerable<DataPoint>(folds[2].TrainSet,
                reuseRowObject: false);

            testSet = mlContext.Data.CreateEnumerable<DataPoint>(folds[2].TestSet,
                reuseRowObject: false);

            PrintPreviewRows(trainSet, testSet);
            // The data in the Train split.
            // [Group, 0], [Features, 0.7262433]
            // [Group, 1], [Features, 0.8173254]
            // [Group, 2], [Features, 0.5588848]
            // [Group, 0], [Features, 0.9060271]
            // [Group, 2], [Features, 0.9775497]
            // [Group, 0], [Features, 0.2737045]
            // 
            // The data in the Test split.
            // [Group, 2], [Features, 0.7680227]
            // [Group, 0], [Features, 0.5581612]
            // [Group, 1], [Features, 0.2060332]
            // [Group, 1], [Features, 0.4421779]
        }

        private static IEnumerable<DataPoint> GenerateRandomDataPoints(int count,
            int seed = 0)

        {
            var random = new Random(seed);
            for (int i = 0; i < count; i++)
            {
                yield return new DataPoint
                {
                    Group = i % 3,

                    // Create random features that are correlated with label.
                    Features = (float)random.NextDouble()
                };
            }
        }

        // Example with features and group column. A data set is a collection of
        // such examples.
        private class DataPoint
        {
            public float Group { get; set; }

            public float Features { get; set; }
        }

        // print helper
        private static void PrintPreviewRows(IEnumerable<DataPoint> trainSet,
            IEnumerable<DataPoint> testSet)

        {

            Console.WriteLine($"The data in the Train split.");
            foreach (var row in trainSet)
                Console.WriteLine($"{row.Group}, {row.Features}");

            Console.WriteLine($"\nThe data in the Test split.");
            foreach (var row in testSet)
                Console.WriteLine($"{row.Group}, {row.Features}");
        }
    }
}