"Resolving incomplete windows" on Moving Average

Dear KNIMErs,

I've run into a nice option in the "Moving Aggregation" node, namely "Resolve missing values for incomplete windows". I suppose this is not yet implemented in the "Moving Average" node?

So far I couldn't find any equivalent, leaving me with gaps in the averaged series when using simple of Gaussian smooths. These gaps can be filled with adaptive window sizes in a loop, but it's quite cumbersome to implement, and not exactly efficient.

Any plans for moving this function across from "Moving Aggregation"? Oh, and dealing with "Moving Average's" occasional "NaN" output gracefully would be nice, too - just like the R nodes do now, going for missing values instead if so desired.


Thanks for the feedback, this is a very good point. I have opened a ticket for you and see what we can do. 

I'm not sure I follow your last point about the R nodes.  Can you elaborate?




Thanks! As for the R nodes, I've been puzzled to see that the Moving Average node occasionally generates "NaN" results. Searching the site for use of the "NaN" keyword, I found the 2.10.0 changelog mentioning the following enhancement:

Enh 5110: R (Interactive): Nodes with table output should have checkbox to represent extremes (NaN & Infinity) as Missing Cells or as-is

It'd be great if we had the same kind of control over the "Moving Average" node, as KNIME isn't really graceful about "NaN" results.