- This topic has 3 replies, 2 voices, and was last updated 7 years ago by
.
- Posts
Hello Sir!
In this past paper, we are told to find the value of year 2013 Q3.
Now, going through this question, it seems that the trend figure had stopped to 2012 Q4. After that, the examiner seems to have made a guess for remaining trend figures.
Don’t we just take the formula y=T+S?
where y would be the actual sales data, T the trend and S the seasonal variation.
We are given the actual data for 2012 Q1 to be 137 and Seasonal variation to be -14.71. Then we can get the trend to be 151.71.Is my method the right approach? If not,can you guide me please?
AND, how do we approach Q3 2013 sales figure when you are not given the trend but seasonal variation is provided (on the analysis of sequence order)? Do we make a guess here on the trend?
Thank you.
The examiner has been arbitrary, but justified it on the basis of
I would have done it by saying that the trend (that’s given by the moving average) rises from 131 to 146.25 in 9 jumps, so (146.25 – 131)/9 per season = 1.69.season.
The last trend figure is 146.25 for 2012, qtr 4. To get to 2013 qtr 3 requires jumping forward 3 quarters from 2012 qtr4 so the predicted trend could be: 146.25 + 3 x 1.69 = 154.8. To this we have to add the seasonal variation of 22.55 to give 177.35.
As the answer says, there is no agreed way to extrapolate the trend and the examiner used a ‘hand drawn’ approach. You couldn’t easily do that in the exam.
Your approach simply reconstructs the trend for figure Q1 2012 but the trend is assumed to increase steadily each quarter.
And this seems to be additive model? How do we tend to know that it is additive model? Is it by the use of + and – signs?
And if we are following multiplicative model, do we take y=T * S approach?
Following this example, would it be like this:[146.25 + (1.69*3)] * 22.55
Thank you.
You can see it’s the additive model from the data given where they calculate seasonal variations by taking differences between the trend and the actual. Multiplicative woud have taken a %.
Your suggested calculation is correct. Find the trend figure then raise/lower it by the variation %.
- You must be logged in to reply to this topic.