Trend lines through moving averages, the mean seasonal effect, and forecasting future values from a time series.

What this dot point is asking

AQA wants you to draw a trend line through the moving averages, calculate the mean seasonal effect, and combine the two to forecast a future value, while recognising the limits of such predictions. This dot point pulls together the whole time series module: the moving averages give the trend, and the seasonal effect restores the season to produce a forecast.

Drawing the trend line

Drawing the line through the smoothed moving averages rather than the jagged raw data is what makes the trend clear: the raw points jump up and down with the season, but the moving averages already sit close to a straight line. As with any line of best fit, balance the points either side, and you can read or extend it to estimate the trend value at any time. The gradient of the trend line is itself meaningful: it gives the average change per time period once the season is removed, so a steeper line means faster long-term growth or decline. Extending the line beyond the data to read a future trend value is the first step of a forecast, but it is also extrapolation, so the further you extend it the less you should trust it.

The seasonal effect

The seasonal effect measures how far a particular season usually sits above or below the trend. A positive mean seasonal effect (say $+7$ for summer) means that season is typically $7$ above the trend; a negative one means it is typically below. Averaging across several years smooths out the random variation so the figure reflects the genuine seasonal pull, not a one-off. In a complete analysis the four quarterly mean seasonal effects should roughly cancel to zero, because the trend already captures the average level.

This is also how you reverse the process to compare seasons. The mean seasonal effects let you say which quarter is consistently the strongest and which the weakest, independent of the overall trend: the quarter with the largest positive mean seasonal effect is the reliable peak. They can also reveal a one-off anomaly: if one quarter's actual seasonal effect is far from its mean, that quarter behaved unusually (perhaps because of a special event), which is worth flagging when you interpret the data. So the seasonal effects do double duty, supporting both forecasting and the description of the seasonal pattern itself.

Forecasting future values

Limitations of forecasts

Forecasting is a form of extrapolation: you are predicting beyond the range of the data. The further ahead you predict, the less reliable it becomes, because the trend may flatten or reverse and the underlying conditions (the economy, competition, fashion) may change. Always note this limitation when asked to evaluate a forecast, because recognising it is part of the marks.