Hi "Knimers"!
I was trying to check if it is possible to do Adstock transformation in my data using Knime.
Anyone tried?
There is an interesting node for moving averages but for Adstock is a bit different.
I've found the material below with explanations about Adstock transformation types:
https://mpra.ub.uni-muenchen.de/7683/4/Adstock_Model.pdf
I appreciate any comment related with this task.
Cheers
Hi Fabio,
sorry, I also never tried this. Maybe you could use the Moving Aggregation node in some way for this?
Cheers, Iris
Hi Iris!
MA is not the same but for my task, I could find a type of MA that worked well.
Tks any way!
Has anyone figured this out? What would be needed is the ability to combine the Moving Aggregation Column a custom aggregate. The way adstock works is that it assumes you retain a certain amount of the previous periods investment/exposure in my attached example, it would be 10%. I tried to figure this out with the lag, but the key is the cumulative nature that gives a decay over time, and that I can't figure out using existing nodes. Has anyone built anything custom for this?
Hi there...despite the fact it was a quite old task, I remember that at that time I asked a friend to help me using R for this task.
Unfortunately, I don't have the code because I´m not working there anymore.
Anyway, is not that complex if you use R or Python