Forecasting Approach

[n_meghanath]

Are you guys aware about the forecasting approach given below,

week 1 actual offered volume=w
week 2 actual offered volume=x
week 3 actual offered volume=y
week 4 actual offered volume=z

Week 5 forecast=???

week 1 actual offered volume=w*a%=e
week 2 actual offered volume=x*b%=f
week 3 actual offered volume=y*c%=g
week 4 actual offered volume=z*d%=h

Now the forecasted volume for week 5 is = e+f+g+h

My question is how would we decide the a%,b%,c% and d%. I would assume that we don't have any standard weightage for the industry in case if this approach is valid.

your input and suggestions would be appreciated.

I think you’re right, but depends the program or project you are forecasting, lets talk about Inbound Customer Service = Billing, you probably have a normal or standard behavior from your customers; now let’s talk about Inbound Telemarketing = Sales, here I suggest see the big picture or the whole picture, even you are an Outsourcer or not, depend how you involve all the areas from your client (internal or external) you have to know about the specific efforts and the expected volumes per effort.

Your formula is accurate for a program with a normal behavior and the % depend the weekly absenteeism, season of the year and very important cycle of your product, service, program or marketing effort. The cycle is vital to know and predict how many, when and how often.

Best regards,

Miguel

[BNVANI]

Hi meghanath,

I was quizzed why your are adding week1,2,3 & 4 for arriving at Week5 volume.

Ideally it has to be average of week1,2,3 & 4.

a%,b%,c% and d% depends ... percentage increase or decrease you are expecting on the base numbers i.e w,x,y and Z

Regards,
Vani

[n_meghanath]

Vani,

I think you are confused and at the same time I admit that I have made a messy expression.

1.The week 5 forecast is not the addition of week 1 to week 4 volume.
2.The methodology is not a point estimates or averaging approach nor Time series.
3. An example to re explain the approach:(Real data)
Actual volume
week1 1468 *3%= 44.04
week2 1410 *13%=183.3
week3 1450 *24%=348
week4 1480 *60%=888
average 1452

Total 100% 1463

The week 5 forecast figure as per the following approach is 1463, and the actual volume which came against this figure in week 5 was 1464. Carrying out the same excercise for last 28 weeks gives >97% forecast accuracy with the above % assumption.


This approach looks very simple and looks like a very simple linear approach however the % (3%,13%......60%) weightage given to derive the next week forecast is still a mystery to me.

Finally: My question is the logical approach of giving 60% in the fourth week (immediate past week), 24% in the 2nd last week and so on......

I am not too sure if this is a pervalent approach ( I use a combination of point estimates, average approach and TIme series) but I have come across few centers who forecast their volume using this approach but the surprising part is some centers take 58% in the fourth week , 30% in the last 2nd week.....

Why 60%? why 58%, why 30%, why24%?

I think I have myself clear now.