Concatenating the positive and negative cases into a single Spark dataframe