Transcription of Finding Informative Features - CMU Statistics
{{id}} {{{paragraph}}}
Finding Informative Features36-350: Data Mining4 September 2009 Readings: David P. Feldman, Introduction to Information Theory , chapter1 ( )Principles of Data Mining, sections , , and I mentioned last time, everything we have learned how to do so far similarity searching, nearest-neighbor and prototype classification, multidimen-sional scaling relies on our having a vector offeaturesorattributesfor eachobject in data set. (The dimensionality of vector space equals the number offeatures.) The success of our procedures depends on our choosing good Features ,but I ve said very little about how to do this. In part this is because designinggood representations inevitably depends on domain knowledge.
Similarly, our uncertainty about the class C, in the absence of any other information, is just the entropy of C: H[C] = X c Pr(C= c)log 2 Pr(C= c) Now suppose we observe the value of the feature X.
Domain:
Source:
Link to this page:
Please notify us if you found a problem with this document:
{{id}} {{{paragraph}}}