Spark custom aggregation : collect_list+UDF vs UDAF

Question

I often have the need to perform custom aggregations on dataframes in spark 2.1, and used these two approaches :

• using groupby/collect_list to get all the valu