"CReW modified outliers" issue

[Ecaterina]

Hi.

I try to use "CReW modified outliers" to normalize the data for Arima and ETS(for prediction), but I could not do this. 

How you can see in the  picture  input,jpg I have just  2 column that I need for the model one is data another is revenueAmount.

I want to normalize the revenueAmount using  "CReW modified outliers" the configuration you can see in the  picture "configuration.jpg"

But the result is strange.

You can see in the picture output.jpg  where is more coloumn that I expected and  the new column is missing. 

But when I save this data in .csv file I have just 3 columns. Date, RevenueAmount, New_RevenueAmount.

Also when I connect to the Arima , I could not select the new column because I don;t see it. 

What I am doing wrong?

How can I normalize the data without using this "CReW modified outliers"

[CristonS · Accepted solution]

hi @Ecaterina, these issues can be difficult without seeing your data, workflow, results, etc, but I think I know what's happening.

One thing to note is that the "new fields" you're seeing are an artifact of the way the macro was built; these are remnants of the original dataset used to populate the macro (you can see it in the Text Input (1) tool if you open the CReW macro).

I can reproduce this when I run the workflow, then toggle from "Remove Outliers" to "Keep Outliers (and update with Mean"), without re-running the workflow. When you run your workflow with the configuration you want to use, the columns return as expected.

Another note when normalizing data for time series, you might lose the ability to forecast outside of that range. Also, since past values become the predictor variables in the time series tools, you could potentially create bias in your data. Just something to be aware of - you know your data better than anyone!

Please let me know if this resolves your issue!