Last month we examined the use of logistic regression for classification, in which the class of
a data point is predicted given training data1. This month, we look at how to evaluate
classifier performance on a test set--data that were not used for training and for which the
true classification is known. Classifiers are commonly evaluated using either a numeric
metric, such as accuracy, or a graphical representation of performance, such as a receiver
operating characteristic (ROC) curve. We will examine some common classifier metrics and …

Points of Significance: Classification evaluation — Penn State Skip to main navigation Skip
to search Skip to main content Penn State Home Penn State Logo Help & FAQ Home
Researchers Research output Research units Equipment Grants & Projects Prizes Activities
Search by expertise, name or affiliation Points of Significance: Classification evaluation
Jake Lever, Martin Krzywinski, Naomi Altman Statistics Research output: Contribution to
journal › Short survey › peer-review 165 Scopus citations Overview Fingerprint Original …