DataMinig_lectures10_11_classification课件.pptVIP

ICS 278: Data MiningLectures 10,11: Classification Algorithms;Notation;Classification;Classification v. Regression;Decision Region Terminlogy;Probabilistic view of Classification;Example of Probabilistic Classification;Example of Probabilistic Classification;Example of Probabilistic Classification;Decision Regions and Bayes Error Rate;Decision Regions and Bayes Error Rate;Procedure for optimal Bayes classifier;3 categories of classifiers in general;Which type of classifier is appropriate?;Example: classifying spam email;Examples of classifiers;Na?ve Bayes Classifiers;Link between Logistic Regression and Na?ve Bayes;Imbalanced Class Distributions;Ranking and Lift Curves; Target variable = response/no-response from mailing campaign Training and test sets each of size 250k Standard model had 80 variables: variable selection reduced this to 7 Note non-monotonicity in lower curve (undesirable);ROC plots;ROC Example;ROC Plot;Example: Link Prediction in Coauthor Graphs O Madadhain, Hutchins, Smyth, SIGKDD, 2005;Evaluation Metrics;Link Prediction Evaluation;Link Prediction Evaluation;Link Prediction Evaluation;Lift Curves for Different Models;Interpretation of Ranking at Top of Ranked List ;Lift Curves for Different Models;Calibration;Calibration in Probabilistic Prediction;Linear Discriminants;Nearest Neighbor Classifiers;Local Decision Boundaries;Finding the Decision Boundaries;Finding the Decision Boundaries;Finding the Decision Boundaries;Overall Boundary = Piecewise Linear;Example: Choosing k in kNN (example from G. Ridgeway, 2003);Decision Tree Classifiers;Decision Tree Example;Decision Tree Example;Decision Tree Example;Decision Tree Example;Decision Tree Example;Binary split selection criteria;How to Choose the Right-Sized Tree?;Choosing a Good Tree for Prediction;Example: Spam Email Classification;和诸学锨痞彝迪俯诈媚刺母衰汝胶傻涉时空臭澳明掇囤溢饿脂庆咒店垄不DataMinig_lectures10_11_classification课件DataMinig_lectures10_11_classification课件;垒亚啸堂摆句自长誉肇存剐厄高谰痢疾男戌享