public static SuggestedTransform GetRemainingFeatures(List <string> newCols, IntermediateColumn[] existingColumns,
Type currentType, bool includeFeaturesOverride)
{
// Pick up existing features columns, if they exist
var featuresColumnsCount = existingColumns.Count(col =>
(col.Purpose == ColumnPurpose.NumericFeature) &&
(col.ColumnName == DefaultColumnNames.Features));
if (includeFeaturesOverride || featuresColumnsCount > 0)
{
newCols.Insert(0, DefaultColumnNames.Features);
}
return(InferenceHelpers.ConcatColumnsIntoOne(newCols, DefaultColumnNames.Features, currentType, true));
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
List <string> textColumnNames =
columns.Where(
column => column.Type.ItemType().IsText() && column.Purpose == ColumnPurpose.TextFeature)
.Select(column => column.ColumnName).ToList();
if ((textColumnNames.Count == 0) ||
(columns.Count(col => col.Purpose == ColumnPurpose.Label) != 1))
{
yield break;
}
//Concat text columns into one.
string concatTextColumnName;
if (textColumnNames.Count > 1)
{
concatTextColumnName = columns[0].GetTempColumnName("TextConcat");
yield return
(InferenceHelpers.ConcatColumnsIntoOne(textColumnNames, concatTextColumnName, GetType(), false));
}
else
{
concatTextColumnName = textColumnNames.First();
}
//Get Unigram + Trichar for text transform on the concatenated text column.
string featureTextColumn = columns[0].GetTempColumnName("FeaturesText");
yield return(InferenceHelpers.TextTransformUnigramTriChar(Env, concatTextColumnName, featureTextColumn));
//Concat text featurized column into feature column.
List <string> featureCols = new List <string>(new[] { featureTextColumn });
if (columns.Any(
col =>
(col.Purpose == ColumnPurpose.NumericFeature) ||
(col.Purpose == ColumnPurpose.CategoricalFeature)))
{
featureCols.Add(DefaultColumnNames.Features);
}
if (!ExcludeFeaturesConcatTransforms)
{
yield return(InferenceHelpers.ConcatColumnsIntoOne(featureCols, DefaultColumnNames.Features, GetType(), true));
}
}
