Issue while using Composite Forecast Profile

[Former Member]

hi!

I am trying to use a composite forecast profile comprising of

1) A univariate profile with 30% weightage

2) A MLR profile with 70% weightage

When I execute the composite forecast in Interactive Planning, the forecasted values under the MLR profile show a constant value across all time buckets. But If I remove the univariate profile and give 100% weightage to the MLR profile, then the forecasted values are in line with what is obtained when I run a MLR forecast profile indenpendently.

To elaborate on the same :

a) if I run the Univariate profile independently, then I obtain the following forecasted values for 6 future months for sales :

Univariate profile forecasts : 22921, 22921, 22921, 22921, 22921,22921

b) if I run the MLR profile independently, then I obtain the following forecasted values for 6 future months for the dependent variable ( sales in this case ) :

MLR profile forecasts : 18745, 19180, 27647, 20392, 22303, 20380

c) if I run the composite profile with 30-70 weightage to Univariate-MLR, then I obtain the following forecasted values for 6 future months for the dependent variable ( sales in this case ) :

Forecast : 22184, 22184, 22184, 22184, 22184,22184

Univariate profile forecasts : 22921, 22921, 22921, 22921, 22921,22921

MLR profile forecasts : 21868, 21868, 21868, 21868, 21868,21868

My query is why is the MLR profile forecast also showing a constant forecast across all 6 buckets ? If I give 100% weightage to the MLR in the composite profile, then I get the forecasted values as mentioned in case b).

Any clue if the behaviour of the MLR profile is not correct in case c).

regards,

Anirudha

[Former Member · Accepted solution]

Hi Anirudha,

Try to implement 1248396 (eventhough it does not seems 100 % relevant)

It has solved similar issues...

Regards

Julien

[Former Member · Accepted solution]

hi, Julien.

Thanks a lot. The issue has been resolved post application of the note suggested by you.

regards,

Anirudha