public bool VerifyPrediction(ref PredictedData state)
{
return(Vector3.Distance(position, state.position) < 0.1f &&
Vector3.Distance(velocity, state.velocity) < 0.1f &&
sprinting == state.sprinting &&
damageTick == state.damageTick);
}
public static void Main()
{
Console.WriteLine("Hello World!");
var mlContext = new MLContext();
var textLoader = mlContext.Data.CreateTextLoader <Passenger>(separatorChar: ',', hasHeader: true);
var assemblyPath = Assembly.GetExecutingAssembly().Location;
var assetsPath = new DirectoryInfo(assemblyPath)?
.Parent?.Parent?.Parent?.Parent?.Parent?
.GetDirectories("assets")
.FirstOrDefault();
if (!(assetsPath?.Exists ?? false))
{
Console.WriteLine($"Expected datasets at {assetsPath.FullName} and found none!");
}
var trainData = textLoader
.Load($@"{assetsPath.FullName}\titanic.train.csv");
var validateData = textLoader
.Load($@"{assetsPath.FullName}\titanic.validate.csv");
var mlPipeline = mlContext.Transforms
.DropColumns(
nameof(Passenger.PassengerId),
nameof(Passenger.Name),
nameof(Passenger.Ticket),
nameof(Passenger.Fare),
nameof(Passenger.Cabin))
.Append(mlContext.Transforms.ReplaceMissingValues(
nameof(Passenger.Age),
nameof(Passenger.Age),
MissingValueReplacingEstimator.ColumnOptions.ReplacementMode.Mean))
.Append(mlContext.Transforms.Categorical.OneHotEncoding(
nameof(Passenger.Gender),
nameof(Passenger.Gender)))
.Append(mlContext.Transforms.Categorical.OneHotEncoding(
nameof(Passenger.Embarked),
nameof(Passenger.Embarked)))
.Append(mlContext.Transforms.Categorical.OneHotEncoding(
nameof(Passenger.PassengerClass),
nameof(Passenger.PassengerClass)))
.Append(mlContext.Transforms.Concatenate(
"Features",
nameof(Passenger.PassengerClass),
nameof(Passenger.Gender),
nameof(Passenger.Age),
nameof(Passenger.SiblingsOrSpouses),
nameof(Passenger.ParentsOrChildren),
nameof(Passenger.Embarked)))
.Append(mlContext.BinaryClassification.Trainers.FastTree(
nameof(Passenger.Survived)))
.Fit(trainData);
var stats = mlContext.BinaryClassification.EvaluateNonCalibrated(
mlPipeline.Transform(trainData),
nameof(OutputModel.Survived));
Console.WriteLine("Training statistics:");
Console.WriteLine($"\tAccuracy: {stats.Accuracy}");
Console.WriteLine($"\tF1: {stats.F1Score}");
var predictor = mlPipeline.CreatePredictionEngine <PredictedData, OutputModel>(mlContext);
foreach (var row in validateData.Preview(2000).RowView)
{
var inputModel = new PredictedData
{
Embarked = row.Values[11].Value.ToString(),
PassengerClass = (float)row.Values[2].Value,
Gender = row.Values[4].Value.ToString(),
Age = (float)row.Values[5].Value,
ParentsOrChildren = (float)row.Values[7].Value,
SiblingsOrSpouses = (float)row.Values[6].Value,
Cabin = row.Values[10].Value.ToString(),
Name = row.Values[3].Value.ToString(),
Fare = (double)row.Values[9].Value,
Ticket = row.Values[8].Value.ToString(),
PassengerId = (int)row.Values[0].Value,
};
var prediction = predictor.Predict(inputModel);
Console.WriteLine($"{inputModel.PassengerId}, {(prediction.Survived ? "1" : "0")}");
}
Console.WriteLine("Press any key to exit...");
Console.ReadKey();
}
