Seasonal data
In cyclical time series, the peaks and troughs tend to occur at irregular intervals. We next consider more regular cycles that are strongly related to the calendar.
In monthly or quarterly data, there is often a pattern of peaks and troughs that repeats in a similar way each year. Daily data often have a similar pattern that repeats each week, and hourly data often have a daily pattern. These are called seasonal patterns.
In seasonal data, it is important that the most recent value is not interpreted in relation to the immediately preceding values. For example, a July unemployment rate may be lower than those in May and June, but this may not indicate a real improvement in the economy if July rates are always lower than those in May and June.
Seasonal patterns make it difficult to assess whether a particular month's value is unusually high or low.
Since seasonal patterns are also likely to occur in the future, it is important to use them for forecasting future values of the series.
International tourists in Hawaii
Many business and economic time series show a strong annual pattern. In particular, data relating to employment, tourism and sales of goods and services often have seasonal peaks and troughs.
The diagram below shows the numbers of international tourists in Hawaii each month from 1990 to 1999.
These figures also exhibit a strong seasonal pattern, though it is hard to identify by eye.
In each year, there are two peaks — one in the winter and another in the summer.
Click on the values to display the figures and verify this seasonal pattern.