By using group by, we get a GroupedData, how could I realize the normalization for each group of data seperatelly? Or for example, now I do something like
val df_list = trans.map(s => {
println(s._1.toString)
val scalerModel = scaler.fit(s._2)
val scaledData = scalerModel.transform(s._2)
scaledData})
where trans is an array of (string, df) and df is dataframe with "features"; I could realize in this way but not very efficient. Is there any better idea?
