Hi Evgeniy, Yes, the reason of the exception is that you are returning an incorrect data type. Flink ML doesn’t have a data type for `DenseVector` but it provides a function called `pyflink.ml.functions.array_to_vector` which returns an `Expression`. So maybe you can modify your UDF to union multiple columns as one column of `DataTypes.ARRAY()`, and then call `pyflink.ml.functions.array_to_vector` on this column.
Best, Xin
