Question 1
1 markAn online shop has data for the number of parcels it sent each quarter from 2021 Q1 to 2024 Q4.
The shop uses a trend line and the mean seasonal variation for Quarter 2 to make two predictions: the number of parcels it will send in 2025 Q2, and the number of parcels it will send in 2028 Q2.
Which of the two predictions is likely to be more reliable?
Give a reason for your answer.
Model answer
The prediction for 2025 Q2 is likely to be more reliable. It is only a year after the end of the data, so the trend and the seasonal pattern are less likely to have changed than they are by 2028.
Explanation
Every time series prediction assumes the trend and the seasonal pattern carry on in the same way. The further into the future the prediction is, the more chance there is that something changes, so predictions close to the end of the data are more reliable. Both predictions here are beyond the data, so 'extrapolation' does not tell them apart.
Mark scheme
-
1 mark
B1 for 2025 Q2, because it is closer to the end of the data, so the trend (and the seasonal pattern) is less likely to have changed by then
oe. The decision must be 2025 Q2 and a reason is needed. Accept 'it is less far into the future than 2028 Q2', 'the further ahead you predict, the more likely the trend is to change', 'it is only one year after the data end, but 2028 Q2 is four years after'. Do not accept: '2025 Q2' with no reason (B0); '2025 Q2 because it is not extrapolation' (B0: both predictions are beyond the data); '2028 Q2 because there will be more data by then' (B0). Ignore any non-contradictory or irrelevant statements.