- This topic has 3 replies, 3 voices, and was last updated 8 years ago by
.
- Posts
Hi Ken,
In BPP Rev kit the above questions a small part of answer was not clear to me .. it says:If the trend value of quarter 1 of 2013 was 148 this would result in a predicted values of 133.3 (148-14.7) . This is just under the actual recorded sale of 137.
Ken , Can you explain how they arrived at the figure of 148 and iam confused between the two figures 148 and 133.3 . Can you please explain it.
The equation of a straight line is y = a + bx, where a = 125·022 and b = 1.84
Quarter 1, 2013, is quarter 13 in the linear regression, so the trend figure predicted would be:
y = 125.022 + 13 x 1.84 = 148.94 (I would have put it at 149 when rounded).
The seasonal adjustment of -14.70 has to be subtracted from that to give the finished predicted value.
Hello sir,
I read the examiner’s answers for this part of the question but I am not able to understand the “time series” section of the examiner’s answers.
Highly appreciate if you could explain me that part which is in examiner’s answers.
Thanks in advance
Moving averages time series analysis divides changes into two parts:
1 The trend – a fairly linear increase or decrease in (here) the sales
2 Seasonal adjustments that can be made to the trend to obtain a more accurate figure for each season.If there are four seasons, as here, the four season moving average for quarters 1 – 4 would centre at time 2.5, half way between 1 and 4. That’s no good as it doe not average on a season. So 1 – 4 seasons are added to 2 – 5 seasons and the total divided by 8 (we have gone from dseason 1 – 5 with 8 seasons in the total) that centres on season 4 (This is well-covered in the OT notes). [(110 + 160 + 155 + 96)+ (160 + 155 + 96 + 116)]8 = 131.
Everything is moved down a season to get the next trend figure etc. The difference between each trend figure and the corresponding actual figure gives the seasonal variation. These are averaged for each season. Thankfully, the work has all been done for you
To predict a value into the future, first predict the trend then superimpose the seasonal variation.
For example, to predict quarter 3 2013:
A way to calculate the trend/season is to say that it goes from 131 to 146.25 in 9 jumps, an average of (146.25 – 131)/9 = 1.7/season
The last trend figure is 146.25 for quarter 4, 2012. Quarter 2, 2013 is 2 quarters further on so the trend figure there could be estimated at 146.25 + 2 x 1.7 = 149.65 [don’t worry that the examiner has 151 – that is based on a hand drawn approach]. The adjust for the seasonal variation for season 3 by adding 22.25.
There’s lots of estimation and guesswork in this subject and many allowable approaches.
- You must be logged in to reply to this topic.