NormalizationCatalog.NormalizeBinning Method

Definition

Overloads

NormalizeBinning(TransformsCatalog, InputOutputColumnPair[], Int64, Boolean, Int32)

Create a NormalizingEstimator, which normalizes by assigning the data into bins with equal density.

NormalizeBinning(TransformsCatalog, String, String, Int64, Boolean, Int32)

Create a NormalizingEstimator, which normalizes by assigning the data into bins with equal density.

NormalizeBinning(TransformsCatalog, InputOutputColumnPair[], Int64, Boolean, Int32)

Create a NormalizingEstimator, which normalizes by assigning the data into bins with equal density.

public static Microsoft.ML.Transforms.NormalizingEstimator NormalizeBinning (this Microsoft.ML.TransformsCatalog catalog, Microsoft.ML.InputOutputColumnPair[] columns, long maximumExampleCount = 1000000000, bool fixZero = true, int maximumBinCount = 1024);
static member NormalizeBinning : Microsoft.ML.TransformsCatalog * Microsoft.ML.InputOutputColumnPair[] * int64 * bool * int -> Microsoft.ML.Transforms.NormalizingEstimator
<Extension()>
Public Function NormalizeBinning (catalog As TransformsCatalog, columns As InputOutputColumnPair(), Optional maximumExampleCount As Long = 1000000000, Optional fixZero As Boolean = true, Optional maximumBinCount As Integer = 1024) As NormalizingEstimator

Parameters

catalog
TransformsCatalog

The transform catalog

columns
InputOutputColumnPair[]

The pairs of input and output columns. The input columns must be of data type Single, Double or a known-sized vector of those types. The data type for the output column will be the same as the associated input column.

maximumExampleCount
Int64

Maximum number of examples used to train the normalizer.

fixZero
Boolean

Whether to map zero to zero, preserving sparsity.

maximumBinCount
Int32

Maximum number of bins (power of 2 recommended).

Returns

Examples

using System;
using System.Collections.Generic;
using System.Collections.Immutable;
using System.Linq;
using Microsoft.ML;
using Microsoft.ML.Data;
using static Microsoft.ML.Transforms.NormalizingTransformer;

namespace Samples.Dynamic
{
    public class NormalizeBinningMulticolumn
    {
        public static void Example()
        {
            // Create a new ML context, for ML.NET operations. It can be used for
            // exception tracking and logging, as well as the source of randomness.
            var mlContext = new MLContext();
            var samples = new List<DataPoint>()
            {
                new DataPoint(){ Features = new float[4] { 8, 1, 3, 0},
                    Features2 = 1 },

                new DataPoint(){ Features = new float[4] { 6, 2, 2, 0},
                    Features2 = 4 },

                new DataPoint(){ Features = new float[4] { 4, 0, 1, 0},
                    Features2 = 1 },

                new DataPoint(){ Features = new float[4] { 2,-1,-1, 1},
                    Features2 = 2 }
            };
            // Convert training data to IDataView, the general data type used in
            // ML.NET.
            var data = mlContext.Data.LoadFromEnumerable(samples);
            // NormalizeBinning normalizes the data by constructing equidensity bins
            // and produce output based on to which bin the original value belongs.
            var normalize = mlContext.Transforms.NormalizeBinning(new[]{
                new InputOutputColumnPair("Features"),
                new InputOutputColumnPair("Features2"),
                },
                maximumBinCount: 4, fixZero: false);

            // Now we can transform the data and look at the output to confirm the
            // behavior of the estimator. This operation doesn't actually evaluate
            // data until we read the data below.
            var normalizeTransform = normalize.Fit(data);
            var transformedData = normalizeTransform.Transform(data);
            var column = transformedData.GetColumn<float[]>("Features").ToArray();
            var column2 = transformedData.GetColumn<float>("Features2").ToArray();

            for (int i = 0; i < column.Length; i++)
                Console.WriteLine(string.Join(", ", column[i].Select(x => x
                .ToString("f4"))) + "\t\t" + column2[i]);
            // Expected output:
            //
            //  Features                            Feature2
            //  1.0000, 0.6667, 1.0000, 0.0000          0
            //  0.6667, 1.0000, 0.6667, 0.0000          1
            //  0.3333, 0.3333, 0.3333, 0.0000          0
            //  0.0000, 0.0000, 0.0000, 1.0000          0.5
        }

        private class DataPoint
        {
            [VectorType(4)]
            public float[] Features { get; set; }

            public float Features2 { get; set; }
        }
    }
}

Applies to

NormalizeBinning(TransformsCatalog, String, String, Int64, Boolean, Int32)

Create a NormalizingEstimator, which normalizes by assigning the data into bins with equal density.

public static Microsoft.ML.Transforms.NormalizingEstimator NormalizeBinning (this Microsoft.ML.TransformsCatalog catalog, string outputColumnName, string inputColumnName = default, long maximumExampleCount = 1000000000, bool fixZero = true, int maximumBinCount = 1024);
static member NormalizeBinning : Microsoft.ML.TransformsCatalog * string * string * int64 * bool * int -> Microsoft.ML.Transforms.NormalizingEstimator
<Extension()>
Public Function NormalizeBinning (catalog As TransformsCatalog, outputColumnName As String, Optional inputColumnName As String = Nothing, Optional maximumExampleCount As Long = 1000000000, Optional fixZero As Boolean = true, Optional maximumBinCount As Integer = 1024) As NormalizingEstimator

Parameters

catalog
TransformsCatalog

The transform catalog

outputColumnName
String

Name of the column resulting from the transformation of inputColumnName. The data type on this column is the same as the input column.

inputColumnName
String

Name of the column to transform. If set to null, the value of the outputColumnName will be used as source. The data type on this column should be Single, Double or a known-sized vector of those types.

maximumExampleCount
Int64

Maximum number of examples used to train the normalizer.

fixZero
Boolean

Whether to map zero to zero, preserving sparsity.

maximumBinCount
Int32

Maximum number of bins (power of 2 recommended).

Returns

Examples

using System;
using System.Collections.Generic;
using System.Collections.Immutable;
using System.Linq;
using Microsoft.ML;
using Microsoft.ML.Data;
using static Microsoft.ML.Transforms.NormalizingTransformer;

namespace Samples.Dynamic
{
    public class NormalizeBinning
    {
        public static void Example()
        {
            // Create a new ML context, for ML.NET operations. It can be used for
            // exception tracking and logging, 
            // as well as the source of randomness.
            var mlContext = new MLContext();
            var samples = new List<DataPoint>()
            {
                new DataPoint(){ Features = new float[4] { 8, 1, 3, 0} },
                new DataPoint(){ Features = new float[4] { 6, 2, 2, 0} },
                new DataPoint(){ Features = new float[4] { 4, 0, 1, 0} },
                new DataPoint(){ Features = new float[4] { 2,-1,-1, 1} }
            };
            // Convert training data to IDataView, the general data type used in
            // ML.NET.
            var data = mlContext.Data.LoadFromEnumerable(samples);
            // NormalizeBinning normalizes the data by constructing equidensity bins
            // and produce output based on 
            // to which bin the original value belongs.
            var normalize = mlContext.Transforms.NormalizeBinning("Features",
                maximumBinCount: 4, fixZero: false);

            // NormalizeBinning normalizes the data by constructing equidensity bins
            // and produce output based on to which bin original value belong but
            // make sure zero values would remain zero after normalization. Helps
            // preserve sparsity.
            var normalizeFixZero = mlContext.Transforms.NormalizeBinning("Features",
                maximumBinCount: 4, fixZero: true);

            // Now we can transform the data and look at the output to confirm the
            // behavior of the estimator. This operation doesn't actually evaluate
            // data until we read the data below.
            var normalizeTransform = normalize.Fit(data);
            var transformedData = normalizeTransform.Transform(data);
            var normalizeFixZeroTransform = normalizeFixZero.Fit(data);
            var fixZeroData = normalizeFixZeroTransform.Transform(data);
            var column = transformedData.GetColumn<float[]>("Features").ToArray();
            foreach (var row in column)
                Console.WriteLine(string.Join(", ", row.Select(x => x.ToString(
                    "f4"))));
            // Expected output:
            //  1.0000, 0.6667, 1.0000, 0.0000
            //  0.6667, 1.0000, 0.6667, 0.0000
            //  0.3333, 0.3333, 0.3333, 0.0000
            //  0.0000, 0.0000, 0.0000, 1.0000

            var columnFixZero = fixZeroData.GetColumn<float[]>("Features")
                .ToArray();

            foreach (var row in columnFixZero)
                Console.WriteLine(string.Join(", ", row.Select(x => x.ToString(
                    "f4"))));
            // Expected output:
            //  1.0000, 0.3333, 1.0000, 0.0000
            //  0.6667, 0.6667, 0.6667, 0.0000
            //  0.3333, 0.0000, 0.3333, 0.0000
            //  0.0000, -0.3333, 0.0000, 1.0000

            // Let's get transformation parameters. Since we work with only one
            // column we need to pass 0 as parameter for
            // GetNormalizerModelParameters. If we have multiple columns
            // transformations we need to pass index of InputOutputColumnPair.
            var transformParams = normalizeTransform.GetNormalizerModelParameters(0)
                as BinNormalizerModelParameters<ImmutableArray<float>>;

            var density = transformParams.Density[0];
            var offset = (transformParams.Offset.Length == 0 ? 0 : transformParams
                .Offset[0]);

            Console.WriteLine($"The 0-index value in resulting array would be " +
                $"produce by: y = (Index(x) / {density}) - {offset}");

            Console.WriteLine("Where Index(x) is the index of the bin to which " +
                "x belongs");

            Console.WriteLine("Bins upper bounds are: " + string.Join(" ",
                transformParams.UpperBounds[0]));
            // Expected output:
            //  The 0-index value in resulting array would be produce by: y = (Index(x) / 3) - 0
            //  Where Index(x) is the index of the bin to which x belongs
            //  Bins upper bounds are: 3 5 7 ∞

            var fixZeroParams = (normalizeFixZeroTransform
                .GetNormalizerModelParameters(0) as BinNormalizerModelParameters<
                ImmutableArray<float>>);

            density = fixZeroParams.Density[1];
            offset = (fixZeroParams.Offset.Length == 0 ? 0 : fixZeroParams
                .Offset[1]);

            Console.WriteLine($"The 0-index value in resulting array would be " +
                $"produce by: y = (Index(x) / {density}) - {offset}");

            Console.WriteLine("Where Index(x) is the index of the bin to which x " +
                "belongs");

            Console.WriteLine("Bins upper bounds are: " + string.Join(" ",
                fixZeroParams.UpperBounds[1]));
            // Expected output:
            //  The 0-index value in resulting array would be produce by: y = (Index(x) / 3) - 0.3333333
            //  Where Index(x) is the index of the bin to which x belongs
            //  Bins upper bounds are: -0.5 0.5 1.5 ∞
        }

        private class DataPoint
        {
            [VectorType(4)]
            public float[] Features { get; set; }
        }
    }
}

Applies to