/// <summary>
/// The main program entry point.
/// </summary>
/// <param name="args">The command line arguments.</param>
static void Main(string[] args)
{
// create a machine learning context
var context = new MLContext();
// load data
Console.WriteLine("Loading data....");
var columnDef = new TextLoader.Column[]
{
new TextLoader.Column(nameof(Digit.PixelValues), DataKind.Single, 1, 784),
new TextLoader.Column("Number", DataKind.Single, 0)
};
var trainDataView = context.Data.LoadFromTextFile(
path: trainDataPath,
columns: columnDef,
hasHeader: true,
separatorChar: ',');
var testDataView = context.Data.LoadFromTextFile(
path: testDataPath,
columns: columnDef,
hasHeader: true,
separatorChar: ',');
// build a training pipeline
// step 1: map the number column to a key value and store in the label column
var pipeline = context.Transforms.Conversion.MapValueToKey(
outputColumnName: "Label",
inputColumnName: "Number",
keyOrdinality: ValueToKeyMappingEstimator.KeyOrdinality.ByValue)
// step 2: concatenate all feature columns
.Append(context.Transforms.Concatenate(
"Features",
nameof(Digit.PixelValues)))
// step 3: cache data to speed up training
.AppendCacheCheckpoint(context)
// step 4: train the model with SDCA
.Append(context.MulticlassClassification.Trainers.SdcaMaximumEntropy(
labelColumnName: "Label",
featureColumnName: "Features"))
// step 5: map the label key value back to a number
.Append(context.Transforms.Conversion.MapKeyToValue(
outputColumnName: "Number",
inputColumnName: "Label"));
// train the model
Console.WriteLine("Training model....");
var model = pipeline.Fit(trainDataView);
// use the model to make predictions on the test data
Console.WriteLine("Evaluating model....");
var predictions = model.Transform(testDataView);
// evaluate the predictions
var metrics = context.MulticlassClassification.Evaluate(
data: predictions,
labelColumnName: "Number",
scoreColumnName: "Score");
// show evaluation metrics
Console.WriteLine($"Evaluation metrics");
Console.WriteLine($" MicroAccuracy: {metrics.MicroAccuracy:0.###}");
Console.WriteLine($" MacroAccuracy: {metrics.MacroAccuracy:0.###}");
Console.WriteLine($" LogLoss: {metrics.LogLoss:#.###}");
Console.WriteLine($" LogLossReduction: {metrics.LogLossReduction:#.###}");
Console.WriteLine();
// grab three digits from the test data
var digits = context.Data.CreateEnumerable <Digit>(testDataView, reuseRowObject: false).ToArray();
var testDigits = new Digit[] { digits[5], digits[16], digits[28], digits[63], digits[129] };
// create a prediction engine
var engine = context.Model.CreatePredictionEngine <Digit, DigitPrediction>(model);
// set up a table to show the predictions
var table = new Table(TableConfiguration.Unicode());
table.AddColumn("Digit");
for (var i = 0; i < 10; i++)
{
table.AddColumn($"P{i}");
}
// predict each test digit
for (var i = 0; i < testDigits.Length; i++)
{
var prediction = engine.Predict(testDigits[i]);
table.AddRow(
testDigits[i].Number,
prediction.Score[0].ToString("P2"),
prediction.Score[1].ToString("P2"),
prediction.Score[2].ToString("P2"),
prediction.Score[3].ToString("P2"),
prediction.Score[4].ToString("P2"),
prediction.Score[5].ToString("P2"),
prediction.Score[6].ToString("P2"),
prediction.Score[7].ToString("P2"),
prediction.Score[8].ToString("P2"),
prediction.Score[9].ToString("P2"));
}
// show results
Console.WriteLine(table.ToString());
Console.ReadKey();
}
/// <summary>
/// The main program entry point.
/// </summary>
/// <param name="args">The command line arguments.</param>
static void Main(string[] args)
{
// create a machine learning context
var context = new MLContext();
// load data
Console.WriteLine("Loading data....");
var dataView = context.Data.LoadFromTextFile(
path: dataPath,
columns: new[]
{
new TextLoader.Column(nameof(Digit.PixelValues), DataKind.Single, 1, 784),
new TextLoader.Column("Number", DataKind.Single, 0)
},
hasHeader: false,
separatorChar: ',');
// split data into a training and test set
var partitions = context.MulticlassClassification.TrainTestSplit(dataView, testFraction: 0.2);
// build a training pipeline
// step 1: concatenate all feature columns
var pipeline = context.Transforms.Concatenate(
DefaultColumnNames.Features,
nameof(Digit.PixelValues))
// step 2: cache data to speed up training
.AppendCacheCheckpoint(context)
// step 3: train the model with SDCA
.Append(context.MulticlassClassification.Trainers.StochasticDualCoordinateAscent(
labelColumnName: "Number",
featureColumnName: DefaultColumnNames.Features));
// train the model
Console.WriteLine("Training model....");
var model = pipeline.Fit(partitions.TrainSet);
// use the model to make predictions on the test data
Console.WriteLine("Evaluating model....");
var predictions = model.Transform(partitions.TestSet);
// evaluate the predictions
var metrics = context.MulticlassClassification.Evaluate(
data: predictions,
label: "Number",
score: DefaultColumnNames.Score);
// show evaluation metrics
Console.WriteLine($"Evaluation metrics");
Console.WriteLine($" MicroAccuracy: {metrics.AccuracyMicro:0.###}");
Console.WriteLine($" MacroAccuracy: {metrics.AccuracyMacro:0.###}");
Console.WriteLine($" LogLoss: {metrics.LogLoss:#.###}");
Console.WriteLine($" LogLossReduction: {metrics.LogLossReduction:#.###}");
Console.WriteLine();
// grab three digits from the data: 2, 7, and 9
var digits = context.Data.CreateEnumerable <Digit>(dataView, reuseRowObject: false).ToArray();
var testDigits = new Digit[] { digits[5], digits[12], digits[20] };
// create a prediction engine
var engine = model.CreatePredictionEngine <Digit, DigitPrediction>(context);
// predict each test digit
for (var i = 0; i < testDigits.Length; i++)
{
var prediction = engine.Predict(testDigits[i]);
// show results
Console.WriteLine($"Predicting test digit {i}...");
for (var j = 0; j < 10; j++)
{
Console.WriteLine($" {j}: {prediction.Score[j]:P2}");
}
Console.WriteLine();
}
Console.ReadKey();
}
