STATS 202 Final: STAT 202 Final Exam Fall 2012
31 Jan 2019
School
Department
Course
Professor
Document Summary
This quiz is closed book but you are allowed two double-sided pages of notes. I understand and accept the stanford university honor code. 1: 1) (true or false) (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) A rule-based classi er is determined by a set of mutually exclusive rules. Generally, the test error for a classi er is higher than its training error. The silhouette statistic is used to measure the quality of a classi er. An roc curve can summarize the performance of any clas- si er. Single linkage hierarchical clustering usually produces more compact (or tight) clusters than complete linkage hierarchical clus- tering. Grubbs" test is an outlier detection method designed to con- trol the probability of false detection of an outlier. Fitting a mixture model is a supervised learning problem. All classi ers use the same measure of impurity to estimate the relevant parameters. 2: 2) below is a table of similarities between ve points.
