I want to calculate medians but GroupBy can only handle mean, sum, mode… as aggregation option. I would like to calculate medians of every six rows among hundreds of rows, plus I have hundreds of columns. I know the Conditional Boxplot node can calculate medians, but probably not for my dataset? Anyone an idea? Thanks in advance!
Martin
Hi,
What you need is the ability to run a calculation of rolling subsets of an ordered dataset. Is you dataset a timeseries for one enitity.
One can run calculations out of the box in some of the more complete RDBMS out on the market with the new “analytic/window” functionality which they have been including (SQL Server, Oracle, etc…).
Some other systems have this ability, in particular some with time series functionailty.
I discussed these topics breifly with some of the people at Knime.
Best regards,
Jay