Evaluating the performance of a classification model
- Accuracy and error rate e.g 92.4% accurate and an error rate of 7.6%.
- ROC Curves (Receiver Operating Characteristics) is a plot of true positive rates against false-positive rates.
A good model vs. bad model is about how the model performs on unseen data. It is like memorize vs. understanding. The ROC curve can tell you that.
Evaluating the performance of a regression model
- RMSE – Root mean square error
The value is interpreted as how far the residuals are from zero or as the average distance between the observed values and the model predictions.