Forecasting

Question 1

4 characteristics from last data that can help us to forecast

Answer 1

Trend (increasing or decreasing and how fast)
Seasonality/periodicity (repeating patterns in time)
Economic cycle (not tied to time)
Correlations with other variables

Question 2

What is the forecasting error

Answer 2

The difference between the actual value and the predicted value

Question 3

What’s the Mean absolute deviation

Answer 3

the average of the absolute deviations in the historical predictions

Question 4

What’s the mean squared error

Answer 4

the average of the squared deviations in the historical predictions

Question 5

How do you find the moving average

Answer 5

The forecast of period t + 1 is equal to the average of the last m observations, where m is a given integer parameter.

Question 6

How does single exponential smoothing differ to moving method to produce a smoothed time series

Answer 6

Whereas in moving averages, the past observations were all treated equally (and completely ignored if they were further away than m time periods), exponential smoothing assigns exponentially decreasing weights as the observations get older.

Question 7

Equation for the one step forecast (smoothed average at time t) when you have a smoothing parameter α

Answer 7

α x [sum i =0 to n] (1-α)^i x X(t-i)

Question 8

When does single exponential smoothing not excel

Answer 8

In following data where there is a trend,

.

Question 9

What characteristic does double exponential smoothing allow for

Answer 9

A trend

Question 10

What characteristics does Winters’ multiplicative smoothing allow for

Answer 10

Trend and seasonality

Question 11

How to adapt multiplicative smoothing equations to winters additive smoothing

Answer 11

Seasonality is used an addition rather than a multiplier

For the first term in the base equation Et (normal single exponential smoothing) and seasonality equations, change the coefficient from division to subtraction.

Question 12

How to get exponential smoothing with trend equations

Answer 12

Enter S=1 into winters multiplicative smoothing equations

Question 13

How to get exponential smoothing equation from multiplicative smoothing equations

Answer 13

Only have a base equation (doesn’t account for trend or seasonality) so these values just plug in as 1

Question 14

How to get exponential smoothing with multiplicative seasonality equations

Answer 14

Enter T=0 into multiplicative smoothing equations (trend isn’t considered.

Question 15

How to get exponential smoothing with additive seasonality equations

Answer 15

Enter T=0 into winters additive method.