3.3.3. Root Relative Square Error

The root relative square error (RRSE) is the ratio of the square root of the sum of the squared errors to the sum of the squared errors of a simple predictor. Again, the simple predictor is the average of the target values. The RRSE is given as

$$RRSE = \sqrt{\frac{\sum\_{i=1}^{n} (P\_i - A\_i)^2}{\sum\_{i=1}^{n} (\bar{A} - A\_i)^2}} \times 100\% \tag{12}$$

where all terms remain as previously defined. By computing the square root of the relative squared error, the RRSE reduces the error to a similar magnitude range as the RAE. However, unlike the RAE, the RRSE penalizes outliers with large error values, thus allowing models with plausible outliers to be easily identified.
