Forecast Holt Winters doesn't work on my data

[claire Banniard]

Hi all,

 

I'm triying to use the module Forecast on some charts but allways get the same error message : A script error occured when calculating Forecast. No more than that, and I can't manage to understand why it doesn't work properly.

In the begining I get errors saying that datas had to be separated by the same space on x axis. I solved the issue.

My chart only represents a cumulated sum over the past year. I have now datas for each day.

My data set in the begining was more complicated than now (with filter columns ) and the cumulated sum was calculated in spotfire.

I tried to simplify the dataset in order to help the Forcast. Buut I allways get the same message.

On my last data set the cumulated sum is calculated before spotfire, spotfire only plot the values for each day. And there is no hole in the datas, every day has a value.

Really I don't understand why it doesn't work.

 

Has anyone experienced this before

[Gaia Paolini]

since it seems to depend on the specific dataset, would you be able to post an anonymised sample of your data (including info on the column datatypes)

[claire Banniard]

Hi,

 

I have added an anonymised version of my dataset.

 

Calendar_date is a date

 

number_cum is an interger (representing days )

 

nb_open_event_cum is an integer, cumulated sum on past year

 

nb_closed_event_cumis an integer, cumulated sum on past year

[Gaia Paolini]

I am not sure I am doing what you were doing. I added a line chart withh calendar_date as X and nb_closed_events_cum as Y. I added a Holt Winter forecast from Lines and Curves. I did not specify any particular input parameter. It does not look great but I do not get an error. Were you following a different process

[claire Banniard]

Hi, we can close this post. After internal discussions it appears that my Spotfire version is not connected to TERR, thats why I recieved a script error message. A collegue has oppened my analysis and the Forcast appears on is version.

Thank you for help.