Forecasting:
(a) All forecasts are never 100% accurate but subject to error.
- How is forecast error calculated?
- Identify and describe three common measures of forecast error. Then illustrate how each is calculated by constructing a 4-period example.
(b) Consider the following table of monthly sales of car tyres by a local company:
|
Month
|
Unit Sales
|
|
January
|
400
|
|
February
|
500
|
|
March
|
540
|
|
April
|
560
|
|
May
|
600
|
|
June
|
?
|
(i) Using a 2-month moving average develop forecasts sales for March to June inclusive.
(ii) Using a 2-month weighted moving average, with weights of 2 for the most recent month and 1 for the previous month develop forecasts sales for March to June inclusive.
(iii) The sales manager had predicted sales for January of 400 units. Using exponential smoothing with a weight of 0.3 develop forecasts sales for March to June inclusive.
(iv) Which of the three techniques gives the most accurate forecasts? How do you know?
(c) Describe the four patterns typically found in time series data. What is meant by the expression "decomposition" with regard to forecasting? Briefly describe the process.