ARIMA - impossible to work with generated forecast from Arima

Moucho84

dear Community,

I did a revenue forecast using the Arima model, before this and to have stationary data, I use the operator logarithm, I obtained the forecast from Arima operator followed by "apply forecast", my next step was to generate an attribute using the function "exp" (exponentiel) however I cannot see the select any attribute except the date ? 

I can see that the attribute is appearing in the results under a new column of data combining historical revenue and forecasted revenue under the name "Revenue and forecast" but impossible to select this attribute. 

can you help me pleae ?

MartinLiebig

Hi,

can't you just use Generate Attributes and use exp([Revenue and Forecast]) there?

BR,

Martin

MartinLiebig

There seems to be some issue with the meta data here. Can you share to process so @David_A and I can check?

Best,

Martin

MartinLiebig

Hi,

what happens if you just type it in and run it? Might be just a meta data issue.

BR,

Martin

Moucho84

Hi Martin,

yes I have the result where I can see
- the revenue for past period (test data)
- the forcasted revenue in a separate column for the horizons required
- the sum of original revenue + the forecasted data under a third column named : Revenue and Forecast

problem is that the data are logarithm, I need to use the operator : generate attribute with « exp » to have the exponential value and the real revenue. However I cannot find this attribute in the process design as if not existing.

thanks in advance

MartinLiebig

Hi,

can't you just use Generate Attributes and use exp([Revenue and Forecast]) there?

BR,

Martin

Moucho84

This is what I tried but :  
- generate attribute connected as next step is mot allowing me to select this new attribute in the formula
- and if I just write the formula as suggested (matching perfectly the name) i have an error message :  potential problem « the attribute Revenue and forecast is unknown »  - I launched  the proces and this is now working thank you !!!

MartinLiebig

There seems to be some issue with the meta data here. Can you share to process so @David_A and I can check?

Best,

Martin

Moucho84

Just an additional question, to generate the forecast I had to split the population by business and activity using « filter example »

now I am dupplicating the process by country, activity and obtain results for each tab (having 6 countries and 6 business line and in it 3 activity - i am ending with 6*6*3 = 108 tabs of results)

Which operator can I use at the end to group all result into 1 tab (combining all country, their busines  and activity with their name) ?

or shall I explore a kind of loop process, allowing the process to do it activity by activity and country by country ?