The definitions and formulas for the statistics of fit that are available in SAS Visual Forecasting are described below. You can use statistics of fit to measure how well different models fit the data. The statistics of fit for the various forecasting models can be printed or stored in a data set.
The formulas below use the notations described in Notations Used in the Formulas.
For most statistics of fit, a lower number indicates a better fit. If a different measure is used to determine a better fit, it is indicated in the description below. Choose from one of the following options.
The adjusted
R2 statistic, , where R2 is defined in the RSQUARE statistic .
Better performance is indicated when this measurement is closer to 1.
Akaike’s information criterion with an
empirical correction for small sample sizes,
Better performance is indicated when this measurement is closer to 1.
The geometric mean of the absolute error as a percentage of the standard deviation
The geometric mean percent error
The geometric mean absolute predictive percent prediction error
The geometric mean of the relative absolute errors
The geometric mean of the absolute symmetric percent errors
The mean of the in-sample absolute scaled errors (IMASE):
The maximum of the absolute error as a percentage of the standard deviation
The largest prediction error
The largest percent prediction error, . The summation ignores observations where
.
The maximum of the predictive percent errors
The maximum of the relative errors
The maximum of the symmetric percent errors
The mean of the absolute error as a percentage of the standard deviation
The mean of the absolute symmetric predictive percent error
The mean of the absolute scaled errors
The mean percent prediction error, . The summation ignores observations where
.
The mean of the predictive percent error
The mean of the relative absolute errors
The mean of the relative errors
The mean squared prediction error calculated from
the one-step-ahead forecasts, . This formula enables you to evaluate small holdout samples.
The mean of the symmetric percent errors
The symmetric mean of the absolute percent error
The median of the absolute error as a percentage of the standard deviation
The median of the percent errors
The median of the predictive percent errors
The median of the relative absolute errors
The median of the symmetric percent errors
The minimum of the absolute error as a percentage of the standard deviation
The smallest prediction error
The smallest percent prediction error, . The summation ignores observations where
.
The smallest predictive percent error
The smallest relative error
The smallest symmetric percent error
The R2 statistic, . If the model fits the series badly, the model sum of squares errors,
SSE, might be larger than SST and the
R2 statistic is negative.
Better performance is indicated when this measurement is closer to 1.
The random walk
R2 statistic (Harvey’s R2 statistic using the random walk model for comparison), , where RWSSE (sum of square errors of the random walk model) is
, and
Better performance is indicated by a higher value.
The root value of the mean of the square scaled errors
The sum of the squared prediction errors, , where
is the one-step predicted value
The unbiased mean squared error
The unbiased root mean squared error
The total sum of squares for the series corrected
for the mean: , where
is the series mean
The total sum of squares for the series,
uncorrected for the mean: