using FLDA to classify IRIS data set

Posted 2011.10.14 01:48


1. To do

Classify IRIS data set using FLDA and 1 nearest neighborhood classifier 


2. Test Result

A raw data
1. using all 4 features 
 94.6%

2. using 2 features (among 4)
 68%

B projected data using FLDA
1. using all 4 features 

a) reduction to d = 4

 93.3%

b) reduction to d = 2

 93.3%

c) reduction to d = 1

    94.6%

2. using 2 features (among 4)

a) reduction to d = 2

 74.7%

b) reduction to d = 1

    77.3% 


 -> FLDA is an useful tool for dimension reduction in pattern classification. 
« PREV : 1 : ··· : 135 : 136 : 137 : 138 : 139 : 140 : 141 : 142 : 143 : NEXT »