Module 05 Activity

Scenario

Your time series charts are technically correct and several of them report the season rather than the trend.

What you build

A time series set with a justified interval, honest smoothing, labelled indexing and forecasts that show their uncertainty.

Steps

1. For each series, state the interval you chose and the decision that justifies it. Try one level coarser and one finer.
2. Identify any seasonal series and replace consecutive-period comparisons with same-period-last-year comparisons.
3. If you smooth anything, report the window, the points lost at the ends, and the lag at a turning point.
4. Index any chart comparing differently-sized series, label the base period on the axis, and put the absolute values somewhere.
5. Redraw any forecast in a distinct style with a shaded interval and a visible mark where the actuals stop.
6. Say which of these changes altered a conclusion.

Evidence to hand in

• The interval decision per series, with the two alternatives tried.
• The seasonal comparison, rewritten.
• The smoothing report with window, lost points and lag.
• The indexed chart with its base period labelled.
• The forecast with its interval.
• Any conclusion that changed.

Review checklist

• The interval is justified by the decision, not by how the data arrives.
• Seasonal series are compared year on year, not to the previous period.
• Any smoothing reports its window, lost endpoints and lag.
• Indexed charts label the base period and give absolute values somewhere.
• Forecasts are visually distinct and show their interval.

Mark complete