public void ColumnDropper_NoMatchingColumnAtRuntime()
{
var dropper = new ColumnDropper
{
ColumnNameToDrop = "DoesNotExist"
};
var cts = new GracefulCancellationTokenSource();
var toProcess = new DataTable();
toProcess.Columns.Add("Column1");
var ex = Assert.Throws <InvalidOperationException>(() => dropper.ProcessPipelineData(toProcess, new ThrowImmediatelyDataLoadEventListener(), cts.Token));
Assert.IsTrue(ex.Message.Contains("does not exist in the supplied data table"));
}
public void ColumnDropper_Successful()
{
var dropper = new ColumnDropper
{
ColumnNameToDrop = "ToDrop"
};
var cts = new GracefulCancellationTokenSource();
var toProcess = new DataTable();
toProcess.Columns.Add("ToDrop");
var processed = dropper.ProcessPipelineData(toProcess, new ThrowImmediatelyDataLoadEventListener(), cts.Token);
Assert.AreEqual(0, processed.Columns.Count);
Assert.AreEqual(false, processed.Columns.Contains("ToDrop"));
}
public static PredictionModel <HousePrice, HousePricePredicted> Train()
{
string _datapath = @"pp-2018.csv";
var pipeline = new LearningPipeline();
var dropper = new ColumnDropper();
pipeline.Add(new TextLoader(_datapath).CreateFrom <HousePrice>(useHeader: false, separator: ','));
pipeline.Add(new ColumnDropper()
{
Column = new string[] { "DateOfTransfer", "Street", "Locality", "PAON", "SAON" }
});
pipeline.Add(new CategoricalOneHotVectorizer(
"PostCode",
"City",
"District",
"RecordStatus",
"Duration",
"PropertyType",
"OldNew"));
pipeline.Add(new ColumnConcatenator(
"Features",
"PostCode",
"City",
"District",
"RecordStatus",
"Duration",
"PropertyType",
"OldNew"));
pipeline.Add(new FastTreeRegressor()
{
LabelColumn = "Price"
});
return(pipeline.Train <HousePrice, HousePricePredicted>());
}
