Former Member

# How system is calculating ExPost Values

Hi Experts,

Could you please help me how below values are getting populating.

History Ex-Post

Jan 2,000

Feb 1,520 2,000

Mar 2,000 1,856

Apr 2,520 1,899

May 1,520 1,899

Jun 2,520 1,899

Jul 1,520 1,865

Aug 2,500 1,762

Sep 1,500 1,791

Oct 2,500 1,704

Nov 1,500 1,943

Dec 2,000 1,810

I am using Forecast model with parameters: Constant model 10

Alpha- 0.30 & Sigma- 1.25, Outlier- Expost

Thanks

Hitesh

##### Add a comment
10|10000 characters needed characters exceeded

### Related questions

• Former Member
Posted on Jun 23, 2009 at 05:00 PM

History -

Ex-Post

Jan -

2,000

Feb----

1,520 -

2,000

Mar----

2,000 -

1,856

Apr----

2,520 -

1,899

May----

1,520 -

1,899

Jun----

2,520 -

1,899

Jul----

1,520 -

1,865

Aug -

2,500 -

1,762

Sep -

1,500 -

1,791

Oct -

2,500 -

1,704

Nov -

1,500 -

1,943

Dec -

2,000 -

1,810

##### Add a comment
10|10000 characters needed characters exceeded
• Former Member
Posted on Jun 24, 2009 at 05:14 AM

Hi Hitesh,

Just gone through following link. You will find the formuleas of diffrent forecast model. In that you will find the how does system calculate expost forecast & other things also.

forecast

Actualy this sheet made by Percx (SDN Member). Now same i am sharing with you.

Link is : APO Basic FC Models xl

Regards

Sujay

Edited by: Sujay Joshi on Jun 24, 2009 10:50 AM

##### Add a comment
10|10000 characters needed characters exceeded
• Former Member

Thanks Sujey,

I have gone through the sheet, and helped me.

Few doubts:

It is caluculating 3 Cycles, what is that and can we customize these cycles to 1.

And while going for next cycle, it is rolling value to period-1, but how it is calculating last period.

As if you check the 1st Ord. Exp Smoothing, and cycle 2, period = -1 Value= 38?

Thanks for your help!

Hitesh

• Former Member
Posted on Aug 11, 2009 at 12:26 PM

I am still working on it.

##### Add a comment
10|10000 characters needed characters exceeded
• Former Member
Posted on Aug 11, 2009 at 02:47 PM

Hi Hitesh,

This question is marked as answered... are you still having issue?

Just in case it looks like you are trying to calculate the ex-post forecast.

Well since you are using a constant model, the formula is quite easy:

you have only a base forecast and only the alpha facvtor is used.

First ex post is for 2nd iteration: it is eaul to the first iteration of history.

In your case 2000

then you use the following formula

expost(n)=history(n-1) * alpha + ex-post(n-1)*(1-alpha)

let's say: E(n)=H(n-1)A+E(n-1)(1-A)

E3=15200.3+20000.7=1856

E4=20000.3+18560.7=1899

This way

E5 should be equal to 2085.

That is were the automatic correct happen:

because aprile value 2500 is outside the boundaries the outlier correct it a give it the value of the ex post: 1899.

Then for sure the expost stay the same (E5=18990.3+18990.7=1899!)

Now how are the boundarie calculated?

If I remeber well based on the MAD: mean absole deviation.

in you case for april

And the deviation=2520-1899=621

So the deviation is bigger than the boundarie: the value is corrected by the outlier!

I hope it is readable!

Thanks and Regards

Julien

##### Add a comment
10|10000 characters needed characters exceeded