170 likes | 262 Views
Algorithms: The Basic Methods Witten – Chapter 4. Charles Tappert Professor of Computer Science School of CSIS, Pace University. 1. Inferring Rudimentary Rules 1R (1-rule) Method. This method tests a single attribute and creates a rule that assigns the most frequent class to that attribute.
E N D
Algorithms: The Basic MethodsWitten – Chapter 4 Charles Tappert Professor of Computer Science School of CSIS, Pace University
1. Inferring Rudimentary Rules1R (1-rule) Method This method tests a single attribute and creates a rule that assigns the most frequent class to that attribute
2. Statistical ModelingNaïve Bayes Method Assumes statistical independence – multiply probabilities
Compare: Example from Naïve Bayes Method 3. Divide-and-Conquer:Construct Decision Trees: ID3 Method
7. Instance-Based Learningk-nearest-neighbor method Non-parametric algorithm
8. Clustering: k-means TechniqueTop down method Specify in advance number of clusters, k Randomly choose k seed points Find the closest points to the seed points Compute the means of points closest to each seed point –> seeds for next iteration Stop when the seed points become stable
Clustering: Hierarchy - DendrogramBottom up method Mary Manfredi dissertation Also, see Witten p 81, p 275-278