public static SuggestedTransform ConcatColumnsIntoOne(List <string> columnNames, string concatColumnName,
Type transformType, bool isNumeric)
{
StringBuilder columnName = new StringBuilder();
columnNames.ForEach(column =>
{
columnName.AppendFormat("{0}", column);
});
string columnsToConcat = string.Join(",", columnNames);
var env = new MLContext();
var input = new ColumnConcatenatingEstimator(env, concatColumnName, columnNames.ToArray());
// Not sure if resulting columns will be numeric or text, since concat can apply to either.
ColumnRoutingStructure.AnnotatedName[] columnsSource =
columnNames.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = isNumeric, Name = c
}).ToArray();
ColumnRoutingStructure.AnnotatedName[] columnsDest =
new[] { new ColumnRoutingStructure.AnnotatedName {
IsNumeric = isNumeric, Name = concatColumnName
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
return(new SuggestedTransform(input, routingStructure));
}
public SuggestedTransform(IEstimator <ITransformer> estimator,
ColumnRoutingStructure routingStructure = null, IDictionary <string, string> properties = null)
{
Estimator = estimator;
RoutingStructure = routingStructure;
Properties = properties;
}
public static SuggestedTransform TextTransform(string srcColumn, string dstColumn, IEstimator <ITransformer> estimator)
{
ColumnRoutingStructure.AnnotatedName[] columnsSource =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = srcColumn
} };
ColumnRoutingStructure.AnnotatedName[] columnsDest =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = dstColumn
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
return(new SuggestedTransform(estimator, routingStructure));
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
var selectedColumns = columns.Where(c => !IgnoreColumn(c.Purpose)).ToArray();
var colList = selectedColumns.Select(c => c.ColumnName).ToArray();
bool allColumnsNumeric = selectedColumns.All(c => c.Purpose == ColumnPurpose.NumericFeature && c.Type.ItemType() != BoolType.Instance);
bool allColumnsNonNumeric = selectedColumns.All(c => c.Purpose != ColumnPurpose.NumericFeature);
if (colList.Length > 0)
{
// Check if column is named features and already numeric
if (colList.Length == 1 && colList[0] == DefaultColumnNames.Features && allColumnsNumeric)
{
yield break;
}
if (!allColumnsNumeric && !allColumnsNonNumeric)
{
yield break;
}
List <string> columnList = new List <string>();
foreach (var column in colList)
{
var columnName = new StringBuilder();
columnName.AppendFormat("{0}", column);
columnList.Add(columnName.ToString());
}
var input = new ColumnConcatenatingEstimator(Env, DefaultColumnNames.Features, columnList.ToArray());
ColumnRoutingStructure.AnnotatedName[] columnsSource =
columnList.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = allColumnsNumeric, Name = c
}).ToArray();
ColumnRoutingStructure.AnnotatedName[] columnsDest =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = allColumnsNumeric, Name = DefaultColumnNames.Features
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
yield return(new SuggestedTransform(input, routingStructure));
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
var columnName = new StringBuilder();
var newColumns = new List <TypeConvertingTransformer.ColumnInfo>();
foreach (var column in columns)
{
if (!column.Type.ItemType().IsBool() || column.Purpose != ColumnPurpose.NumericFeature)
{
continue;
}
columnName.AppendFormat("{0}", column.ColumnName);
newColumns.Add(new TypeConvertingTransformer.ColumnInfo(columnName.ToString(),
columnName.ToString(), DataKind.R4));
}
if (columnName.Length > 0)
{
var input = new TypeConvertingEstimator(Env, newColumns.ToArray());
ColumnRoutingStructure.AnnotatedName[] columnsSource =
newColumns.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = c.Input
}).ToArray();
ColumnRoutingStructure.AnnotatedName[] columnsDest =
newColumns.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = c.Output
}).ToArray();
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
yield return(new SuggestedTransform(input, routingStructure));
// Concat featurized columns into existing feature column, if transformed at least one column.
if (!ExcludeFeaturesConcatTransforms)
{
yield return(InferenceHelpers.GetRemainingFeatures(newColumns.Select(c => c.Output).ToList(),
columns, GetType(), IncludeFeaturesOverride));
IncludeFeaturesOverride = true;
}
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
var featureCols = new List <string>();
foreach (var column in columns)
{
if (!column.Type.ItemType().IsText() || column.Purpose != ColumnPurpose.TextFeature)
{
continue;
}
var columnDestSuffix = "_tf";
var columnNameSafe = column.ColumnName;
string columnDestRenamed = $"{columnNameSafe}{columnDestSuffix}";
featureCols.Add(columnDestRenamed);
var input = new TextFeaturizingEstimator(Env, columnDestRenamed, columnNameSafe);
ColumnRoutingStructure.AnnotatedName[] columnsSource =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = columnNameSafe
} };
ColumnRoutingStructure.AnnotatedName[] columnsDest =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = columnDestRenamed
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
yield return(new SuggestedTransform(input, routingStructure));
}
// Concat text featurized columns into existing feature column, if transformed at least one column.
if (!ExcludeFeaturesConcatTransforms && featureCols.Count > 0)
{
yield return(InferenceHelpers.GetRemainingFeatures(featureCols, columns, GetType(), IncludeFeaturesOverride));
IncludeFeaturesOverride = true;
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
bool found = false;
var columnName = new StringBuilder();
foreach (var column in columns)
{
if (column.Type.ItemType() != NumberType.R4 || column.Purpose != ColumnPurpose.NumericFeature)
{
continue;
}
if (!column.HasMissing)
{
continue;
}
found = true;
columnName.AppendFormat("{0}", column.ColumnName);
}
if (found)
{
string name = columnName.ToString();
var input = new MissingValueIndicatorEstimator(Env, name, name);
ColumnRoutingStructure.AnnotatedName[] columnsSource =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = name
} };
ColumnRoutingStructure.AnnotatedName[] columnsDest =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = name
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
yield return(new SuggestedTransform(input, routingStructure));
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
int count = 0;
bool isAllText = true;
var colSpec = new StringBuilder();
var colSpecTextOnly = new List <string>();
var columnList = new List <string>();
foreach (var column in columns)
{
var columnName = new StringBuilder();
if (column.Purpose != ColumnPurpose.Name)
{
continue;
}
count++;
if (colSpec.Length > 0)
{
colSpec.Append(",");
}
colSpec.Append(column.ColumnName);
columnName.Append(column.ColumnName);
columnList.Add(columnName.ToString());
if (column.Type.ItemType().IsText())
{
colSpecTextOnly.Add(column.ColumnName);
}
isAllText = isAllText && column.Type.ItemType().IsText();
}
if (count == 1 && colSpec.ToString() != DefaultColumnNames.Name)
{
var columnName = new StringBuilder();
columnName.AppendFormat("{0}", colSpec);
var input = new ColumnCopyingEstimator(Env, columnName.ToString(), DefaultColumnNames.Name);
ColumnRoutingStructure.AnnotatedName[] columnsSource =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = columnName.ToString()
} };
ColumnRoutingStructure.AnnotatedName[] columnsDest =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = DefaultColumnNames.Name
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
yield return(new SuggestedTransform(input, routingStructure));
}
else if (count > 1)
{
if (string.IsNullOrWhiteSpace(colSpecTextOnly.ToString()))
{
yield break;
}
// suggested grouping name columns into one vector
var input = new ColumnConcatenatingEstimator(Env, DefaultColumnNames.Name, columnList.ToArray());
ColumnRoutingStructure.AnnotatedName[] columnsSource =
columnList.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = c
}).ToArray();
ColumnRoutingStructure.AnnotatedName[] columnsDest =
{ new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = DefaultColumnNames.Name
} };
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
yield return(new SuggestedTransform(input, routingStructure));
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
bool foundCat = false;
bool foundCatHash = false;
var catColumnsNew = new List <OneHotEncodingEstimator.ColumnInfo>();
var catHashColumnsNew = new List <OneHotHashEncodingEstimator.ColumnInfo>();
var featureCols = new List <string>();
foreach (var column in columns)
{
if (!column.Type.ItemType().IsText() || column.Purpose != ColumnPurpose.CategoricalFeature)
{
continue;
}
var columnName = new StringBuilder();
columnName.AppendFormat("{0}", column.ColumnName);
if (IsDictionaryOk(column, EstimatedSampleFraction))
{
foundCat = true;
catColumnsNew.Add(new OneHotEncodingEstimator.ColumnInfo(columnName.ToString(), columnName.ToString()));
}
else
{
foundCatHash = true;
catHashColumnsNew.Add(new OneHotHashEncodingEstimator.ColumnInfo(columnName.ToString(), columnName.ToString()));
}
}
if (foundCat)
{
ColumnRoutingStructure.AnnotatedName[] columnsSource =
catColumnsNew.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = c.Output
}).ToArray();
ColumnRoutingStructure.AnnotatedName[] columnsDest =
catColumnsNew.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = c.Output
}).ToArray();
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
var input = new OneHotEncodingEstimator(Env, catColumnsNew.ToArray());
featureCols.AddRange(catColumnsNew.Select(c => c.Output));
yield return(new SuggestedTransform(input, routingStructure));
}
if (foundCatHash)
{
ColumnRoutingStructure.AnnotatedName[] columnsSource =
catHashColumnsNew.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = c.HashInfo.Output
}).ToArray();
ColumnRoutingStructure.AnnotatedName[] columnsDest =
catHashColumnsNew.Select(c => new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = c.HashInfo.Output
}).ToArray();
var routingStructure = new ColumnRoutingStructure(columnsSource, columnsDest);
var input = new OneHotHashEncodingEstimator(Env, catHashColumnsNew.ToArray());
yield return(new SuggestedTransform(input, routingStructure));
}
if (!ExcludeFeaturesConcatTransforms && featureCols.Count > 0)
{
yield return(InferenceHelpers.GetRemainingFeatures(featureCols, columns, GetType(), IncludeFeaturesOverride));
IncludeFeaturesOverride = true;
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
var firstGroupColId = Array.FindIndex(columns, x => x.Purpose == ColumnPurpose.Group);
if (firstGroupColId < 0)
{
yield break;
}
var col = columns[firstGroupColId];
var columnName = new StringBuilder();
columnName.AppendFormat("{0}", col.ColumnName);
if (col.Type.IsText())
{
// REVIEW: we could potentially apply HashJoin to vectors of text.
string dest = DefaultColumnNames.GroupId;
string source = columnName.ToString();
var input = new OneHotHashEncodingEstimator(Env, new OneHotHashEncodingEstimator.ColumnInfo(dest, source));
var routingStructure = new ColumnRoutingStructure(
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = source
}
},
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = dest
}
}
);
string[] outputColNames = new string[] { DefaultColumnNames.GroupId };
yield return(new SuggestedTransform(input, routingStructure));
}
else if (col.ColumnName != DefaultColumnNames.GroupId)
{
string dest = DefaultColumnNames.GroupId;
string source = columnName.ToString();
var input = new ColumnCopyingEstimator(Env, source, dest);
var routingStructure = new ColumnRoutingStructure(
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = source
}
},
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = dest
}
}
);
yield return(new SuggestedTransform(input, routingStructure));
}
}
public override IEnumerable <SuggestedTransform> Apply(IntermediateColumn[] columns)
{
var lastLabelColId = Array.FindLastIndex(columns, x => x.Purpose == ColumnPurpose.Label);
if (lastLabelColId < 0)
{
yield break;
}
var col = columns[lastLabelColId];
var columnName = new StringBuilder();
columnName.Append(col.ColumnName);
if (col.Type.IsText())
{
col.GetUniqueValueCounts <ReadOnlyMemory <char> >(out var unique, out var _, out var _);
string dest = DefaultColumnNames.Label;
string source = columnName.ToString();
var input = new ValueToKeyMappingEstimator(Env, source, dest);
var routingStructure = new ColumnRoutingStructure(
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = false, Name = source
}
},
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = dest
}
}
);
yield return(new SuggestedTransform(input, routingStructure));
}
else if (col.ColumnName != DefaultColumnNames.Label)
{
string dest = DefaultColumnNames.Label;
string source = columnName.ToString();
var input = new ColumnCopyingEstimator(Env, source, dest);
var routingStructure = new ColumnRoutingStructure(
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = source
}
},
new[]
{
new ColumnRoutingStructure.AnnotatedName {
IsNumeric = true, Name = dest
}
}
);
yield return(new SuggestedTransform(input, routingStructure));
}
}
