Order Statistics
March 5, 2007


Fisher Linear Discrminant Weights

We expect that the closest interpoint distance (the first order statistic) to perform the best in the discrimination task of distinguishing point data sets generated under the Null hypothesis from those generated under the Alternative hypothesis. Then the successive order statistics to perform in an increasing degrading manner. From this naive point of view we expect that the Fisher weights will all be positive, with the highest weights for the first order statistic. The table below shows the Fisher weights to four significant places. They are not all positive!

Fisher Linear Discriminant Weights
Order Statistic Fisher Weight
1 10.84
224.64
3-12.28
424.06
5-3.013
610.99
71.044
87.924
911.99
105.416
117.297
124.768
133.440
14.5212
15-1.064
16-.1111
17-2.200
18-1.457
19-.3259

< Previous >


About
Purpose
Book List
Policy
Experiments
Papers
Torah Code Tutorial
Torah Code Tables
E-Mail
 
Kabbalah Torah

 
| About | Purpose | Book List | Policy | Experiments | Papers | E-Mail |
| Kabbalah Torah |


Last modified Wed, Jan-21-2009, 14:43 MST