Returns a weighted block-coordinate descent model using least squares NOTE: This function assumes that the trainingFeatures have been partitioned by their class index.
Returns a weighted block-coordinate descent model using least squares NOTE: This function assumes that the trainingFeatures have been partitioned by their class index. i.e. each partition of training data contains data for a single class
NOTE: This function makes multiple passes over the training data. Caching @trainingFeatures and @trainingLabels before calling this function is recommended.
Blocks of training data RDDs
training labels RDD
number of passes of co-ordinate descent to run
regularization parameter
how much should positive samples be weighted