Can't understand BI Dev Studio Viewer for MS Time Series

Question

I am analysing Time Series data for sales in customer centers. And my center in Egypt has data since 2005 (presented in the picture below), but the chart in BI Dev Studio is displaying from 200702. I can't understand this, since named query that I use in the data source view is returning the expected result, also when I change the name query only to have data for Egypt customer center I get the expected line that starts from 200501.

Can anyone help me understand this.

I also need some articles or white papers for predicting with MS time series algorithm, and how to test mining model accuracy, please recommend me some.

Answer 1

Hi klimentina,

Are the timestamps synchronized for the two series? Are there any missing time slices for the Egypt series? Can you share the results of the named query that returns both series?

thanks,
Jerry

Answer 2

Hi Jerry,

Thanks for you replay.

I have 69 time slots for Argentina and 44 slots for Egypt. There are a lot of months missing for Egypt in period from (200501 till 201009 last month of sales data, missing like not existing at all, there no records at all ). But i have set MISSING_VALUE_SUBSTITUTION=0 and I expect that this values will be filled with 0 in the time series or missing value substitution will do the job for record of type ([Time Key],[Seria Key],NULL)?

 

---

Klimentina