Data mining Classification--discriminant model----Support vector machine

Support Vector machine is a very popular learning algorithm called SVM for linear and nonlinear data it is the use of a nonlinear transformation, the original training data mapped to high-dimensional space.

At present, it has been widely used in handwritten numeral recognition object recognition and Reference time series prediction Test.

The goal of SVM is to look for a hyper plane, which is concerned with allowing the nearest point of the plane to have the maximum spacing.

The classification mode function is rewritten by the relation of geometric edge and function edge, and finally it is further rewritten into a typical two-time programming problem, after solving, the maximal edge super-plane can be rewritten into decision-making boundary by Lagrange formula ...

SVM is a relatively new concept, is a deterministic algorithm, with good generalization characteristics, using two-time programming techniques for bulk learning, you can use kernel functions to learn complex functions.

---------------------------------

Association classification is based on association rules.

Divided into 2 steps: The first step is the frequent itemsets mining search for the pattern of property-value pairs that recur in the dataset, where each attribute-value pair is considered an item, and multiple attribute-value pairs form frequent itemsets.

Second, rule generation, analysis of frequent itemsets in order to produce association rules.

Estimating the error rate true true negative false positive false negative can be used to evaluate the costs and benefits associated with the classifier model

Data mining Classification--discriminant model----Support vector machine