The sample Pearson correlation coefficient, or 'r': $$r = \frac{\sum ^n _{i=1}(X_i - \bar{X})(Y_i - \bar{Y})}{\sqrt{\sum ^n _{i=1}(X_i - \bar{X})^2} \sqrt{\sum ^n _{i=1}(Y_i - \bar{Y})^2}}$$ It measures the correlation (linear dependence) between the actual predictions and the model's predictions, giving a value between +1 and ?1 inclusive. See: http://en.wikipedia.org/wiki/Pearson_product-moment_correlation_coefficient
| Minimum value | -1 |
| Maximum value | 1 |
| Unit | |
| Optimization | Higher is better |