Statistical Data Mining and Machine Learning Hilary …英文精品课件.pdf

Statistical Data Mining and Machine Learning Hilary Term 2016 Dino Sejdinovic Department of Statistics Oxford Slides and other materials available at: http://www.stats.ox.ac.uk/~sejdinov/sdmml Decision Trees Decision Trees Decision Trees Decision Trees Decision Trees Classification and Regression Trees (CART) Denote input domain by and let the output domain be 1 K (classification) or (regression). A decision tree gives a partition of into R disjoint sets (regions) , such that the fitted decision function is constant on 1 R each region ,j 1 R , i.e. j f x x tree j j Main strengths: easy to use, easy to interpret. Often serve as a starting point for powerful model combination and ensemble techniques: bagging, boosting (random forests). Decision Trees Decision Trees 108 CHAPT Example: NHS Direct Self-help Guide Decision Trees Decision Trees Example: NHS Direct Self-help Guide the NHS Direct self-hel uide left and corres ondin decision tree Decision Trees Decision Trees Decision Trees A decision tree is a hierarchically organized structure, with each node splitting the data space into regions based on value of a single feature (attribute). Some terminology: Paren