The total sum of squares (TSS) is the total variability in the data. Like the [[sample variance]] but without the denominator. $TSS = \sum_{i=1}^n(y_i - \bar y)^2$ Note that the $TSS = RSS + ESS$