In PySpark, I would like to add columns of rolling sum X through the time series date while keep all date available even though some of them are zero.
Original datafr
