Statistics for Data Science

Evaluation

In statistics, the term performance is usually interchangeably with the idea of a model's accuracy. When speaking about a non-statistical database or model, performance is perhaps all about speed—how long it takes for a query to return values, how long it takes to commit a transaction, and so on—and accuracy, and usually revolves around the idea of quality assurance, not an ability to predict values!

When assessing statistical models, a data scientist will look at a model's error rate, the number of misclassifications (that were made by a model), the ratio of the number of correctly predicted compared to the total number of predicted instances (by the model), and more. Again, all of this is dependent on the statistical model type ...