# Discriminant Analysis

Disciriminant analysis is the classical method to solve
Pattern Recognition.

In this method the values of
average and standard deviation
is important.
And
normal distribution
is used as the image of the data.
So if the central points of groups have the meaning, this method is suitable.

There are two ways to solve the discrimination.

##
By a Line

The idea uses a line as the boundary of categories.

###
Difference between Discriminant Analysis and Support Vector Machine (SVM)

Support Vector Machine
is youger method using a line to solve problem.

Discriminant Analysis finds the best line to separate the **distributions** clearly.

Support Vector Machine
find the best line to separate the **data near the line** clearly.

##
By Distances

Mahalanobis' Distance
is used to calculate the distance between the new data and each central point of the groups.

The idea, "distance from the central point" uses the image of normal distribution.

##
Calculation of Probability

This approach can calculate the probability of the categories of the new data.

###
Difference from Logistic Regression Analysis

We can also calculate the probability by
Logistic Regression Analysis
.
By Logistic Regression Analysis, the sum of probability of two category on the same point is 1.

But by Discriminant Analysis, the sum of probability of two category on the same point is not 1.

MT method

Difference of Good Distribution by Methods

Difference of Output by Methods

NEXT Logistic Regression Analysis