FastTreeBinaryFeaturizationEstimator.Options Class

The name of feature column in the IDataView when calling Fit(IDataView). The column type must be a vector of Single. The column called InputColumnName would be mapped to columns called TreesColumnName, LeavesColumnName, and PathsColumnName in the output of TreeEnsembleFeaturizationEstimatorBase and its derived classes. Note that Microsoft.ML.Trainers.FastTree.TreeEnsembleFeaturizationEstimatorBase.FeatureColumnName is not necessary to be the same as the feature column used to train the underlying tree model.

The 0-1 encoding of all leaf nodes' IDs. Its type is a vector of Single. If the given feature vector falls into the first leaf of the first tree, the first element in the 0-1 encoding would be 1. If LeavesColumnName is null, this output column may not be generated.

The 0-1 encoding of the paths to the leaves. If the path to the first tree's leaf is node 1 (2nd node in the first tree), node 3 (4th node in the first tree), and node 5 (6th node in the first tree), the 2nd, 4th, and 6th element in that encoding would be 1. If PathsColumnName is null, this output column may not be generated.

The configuration of FastTreeBinaryTrainer used to train the underlying TreeEnsembleModelParameters.

The name of the column that stores the prediction values of all trees. Its type is a vector of Single and the i-th vector element is the prediction value predicted by the i-th tree. If TreesColumnName is null, this output column may not be generated.