Long time no see, sorry - I’m awfully busy lately. However, I’ve been using KNIME quite intensively again the last days, and now I’m missing a feature. Shouldn’t it be possible to select the “Max” (or “Min”) option for date values in the “Group by” node?
Having said that, a possibility of calculating the range of any numeric variable, or the days between in case of date variables, would be most useful as well.
Any comments of the ease of implementing this, particularly the first request?
I also would love to see Median and MAD (median absolute deviation) as aggregation methods. I could even implement it myself and give you (@knime.org) the patch if you would be willing to integrate it.
Thanks for the support! I fully concur with your median request, but in that case we should maybe also be talking about general quartiles and/or percentiles? Ideally injectable via workflow variables, of course…
we are currently working on this issue. The next version of the group by node will be more flexible in case of data types and extendable. It will give you the possibility to register your own aggregation methods that work with any type of data. We will also include some date related aggregation methods.
Wonderful news! The only question that remains, as always: By when can we expect that? Would a request for a support contract speed up matters?
the new version of KNIME is available and features a new exctension point for your own AggregationMethods. All methods that implement this extension point will be available in the grouopby node and all future nodes that need aggregation mechanism. New methods for date and time have been added as well and the min max is available for all DataValues using the DataValue specific comparator.
We hope you like the extensions.
I do indeed, many thanks! :-)