Complete the following steps to specify the column of data that you want to analyze.
In Series, enter a column of numeric data that were collected at regular intervals and recorded in time order.
(Optional) Select Fit seasonal model. In Period, enter the length of the seasonal pattern. For example, if you collect data monthly and the data have a yearly pattern, enter 12.
If you do not know the seasonal length, use Stat > Time Series > Time Series Plot or Stat > Time Series > Autocorrelation to help you identify the length.
Enter the terms for the model. You must enter at least one term. The seasonal terms are active only if you select Fit seasonal model. For each term, the maximum value you can enter is 5.
Enter a nonseasonal AR term (p). This term is the number of previous values (lags) that affect the current value. If you don't know the number, try the following:
If no lags are significantly correlated, enter 0.
If lags are significantly correlated, enter 1 or 2.
Use autocorrelation to determine whether lags are significantly correlated.
Enter a seasonal AR term (P). This term is the number of lags from the previous season that are significantly correlated with the current season. 1 is sufficient for most seasonal patterns. Use autocorrelation to determine whether lags from the previous season are significantly correlated.
Enter a difference term (d). This value indicates the number of times you subtract the previous data value from the current data value. If you don't know the number, try the following:
If the data do not have a trend and the mean is relatively constant, enter 0.
If the trend is linear or if the mean is not constant, enter 1.
If there is a trend that is not constant and linear, enter 2.
To identify the trend, use Stat > Time Series > Time Series Plot.
Enter a seasonal difference term (D) if the data exhibits a seasonal pattern. 1 is sufficient for most seasonal patterns.
Enter the nonseasonal MA term (q). This term is the number of previous error terms (lags of the forecast errors) that affect the current value. To determine the nonseasonal MA term, look at the partial autocorrelation function. For more information, to go partial autocorrelation function.
Enter a seasonal MA term (Q) if the data exhibits a seasonal pattern. 1 is sufficient for most seasonal patterns.
In this worksheet, Sales contains the number of computers that are sold each month.
Include constant term in model
Select this option to include a constant term if either of the following are true:
The number of differences is 0.
The number of differences is 1, and the data have a trend.
For other cases, you usually do not include the constant term.
Starting values for coefficients
Do not select this option to have Minitab use the default values. Usually, you should try the default values first. If the default values don't produce a solution, enter the column that contains alternative starting values in Starting values for coefficients. The alternative starting values can be the following:
Values that you stored from a previous ARIMA analysis.
Values that you enter for each parameter in the order that they appear in the model, with the constant first.