Flashcards · 4040 · Topic 10

Time series flashcards

12 cards on Topic 10, subtopics 10.1 to 10.2. Pick a subtopic or study them all.

Study mode needs JavaScript. Every card is listed below, so you can still cover the answers and test yourself.

All 12 cards

For a quick read-through or printing. Cover the right-hand side to test yourself.

What is a time series?: Values of one quantity recorded at regular intervals, such as every quarter or every school day, and plotted in time order.
What is a trend?: The long-term direction of the series (rising, falling or flat) once short repeating ups and downs are removed.
What is a moving average?: The mean of a fixed number of consecutive values, moved along the series one value at a time.
How do you choose the number of points in a moving average?: From the length of one cycle, not the number of values: 4 for quarterly data, 5 for school days, 7 for days of the week, 12 for monthly data with a yearly pattern.
When are moving averages centred, and where does the first centred 4-point average belong?: Centre when the number of points is even, by averaging neighbouring pairs. The first centred 4-point average belongs to the third period of the data.
Find the four-point moving averages of 30, 18, 22, 34, 32, 20, 24, 36: 26.0, 26.5, 27.0, 27.5, 28.0 (for example (30 + 18 + 22 + 34) ÷ 4 = 26.0). They rise steadily, so the trend is upward.

What is seasonal variation in a time series?: A regular swing around the trend that returns after the same number of periods each time. A "season" can be a quarter, a day of the week or a month.
How do you calculate the mean seasonal variation (seasonal component)?: For every period with a moving average, find actual value − moving average. Group the differences by season and take the mean of each group.
What is the difference between a seasonal component and a difference?: A difference belongs to one period; the component is the mean of the differences for that season.
Three seasonal components are +5.2, −3.1 and −6.0. What do you expect the fourth to be?: About +3.9, so that the four components total roughly zero.
How do you predict a future value using the trend line and seasonal component?: Read (or extend) the trend line to find the trend value for that period, then add the seasonal component. Add it to a trend value, not to the last actual value.
What assumptions make a time series prediction unreliable?: It assumes the trend continues in a straight line and each season keeps the same effect. A trend line drawn by eye varies, and one-off events are not in the model. The further ahead, the less reliable.