To use this function, choose .
Calculates row-by-row differences between numeric values in a column. You specify the lag (number of rows); Minitab subtracts the value in current row from the value in the lag row.
DIFFERENCES(number, [lag])
In number, specify the column. Minitab subtracts from each row the element lag rows above, and stores the differences in a new column. If you don't specify a value for lag, the differences are calculated between consecutive rows (lag = 1). The first lag rows of the new column will contain the missing value symbol *.
A store manager records the number of customer complaints each month. To monitor monthly variation the manager wants to determine the row-by-row differences in the number of complaints. In , the manager enters the expression DIFFERENCES(C1,1) and stores the results in C2.
C1 | C2 |
---|---|
Complaints | Variation |
55 | * |
45 | −10 |
51 | 6 |
38 | −13 |
59 | 21 |
47 | −12 |
60 | 13 |
49 | −11 |
35 | −14 |
28 | −7 |
Calculating differences can also be an important step in finding trends or seasonality that may affect the data. For example, a toy store company wants to use a times series (ARIMA model) to predict monthly sales. However, they suspect a strong seasonal impact on monthly sales; toy sales in December are always higher than other months due to the holiday season.
To make the data more stationary and help reveal any underlying seasonal pattern before choosing a model, they first calculate differences in monthly sales using a lag of 12. Then they assess autocorrelation of the differences using .