Updating input for algorithm train/test functions to be a Spark SQL RDD instead of a path to data. This should hopefully be helpful for both integration with datasets in different formats and speed up analysis that runs the algorithm over the same dataset multiple times.
Updating input for algorithm train/test functions to be a Spark SQL RDD instead of a path to data. This should hopefully be helpful for both integration with datasets in different formats and speed up analysis that runs the algorithm over the same dataset multiple times.