| | CSIS 405 |
|
|
| | Chapter 2: The Forecast Process, Data Considerations, and Model Selection |
|
|
| | 1. Use ACF to identify the data pattern: |
| | | a. A stationary series: the value of ACF diminishes rapidly (drops after |
| | | the second or third time lag) toward zero as k increases |
|
|
| | | b. A series with the trend (a non-stationary series): the value of ACF |
| | | declines toward zero slowly |
|
|
| | | c. A series with seasonality: ACF (4, 8, …) is significant for quarterly |
| | | data and ACF (12, 24, …..) is significant for monthly data |
|
|
| | | d. A random series: ACF for all lags are not significantly different from |
| | | zero |
|
|
| | 2. Homework: Exercises 3, 8, 9, 10, and 11 (Please use Forecast X for this exercise) |
|
|
| Exercise 8: |
| | | 8b. To obtain autocorrelation using Forecast X: |
|
|
| | | | First, highlight "Year" and "Larceny Thefts" data > click "Add-Ins" at the top > click Forecast X > chick "Data Capture" inside the Forecast X dialog box |
| | | | Make sure "Data is Organized In" "Columns" |
|
|
| | Inside "Data Set" check "Contain Dates" > select "Annual" for Periodicity and "1" for Labels |
|
|
| | | Click "Forecast Method" next to "Data Capture" > click "Analyze" and you will have a 12-period plot of autocorrelation function (ACF) |
Zola_17
HintsforExercise21.xls
