dspss.vef.gov

[klinmy]

i would like to ask sth related to PCA, princomp. i was told tht if we want to reduce the dimension of data, lets say i need to meausre the beautifulness of a house, i came out with 20 principles

eg:

House Principle1(roof) Principle2(color) ... Principle20(material)

H1 30 10 50

H2 10 50 100

H3 80 90 95

by using PCA, princomp in matlab, i can be able to cut off those principles which has less effect on the total measurement. eg, from 20 principles to 10 most influence principles.

i've tried with sth like ths,

[coefs, scores, variences, t2] = princomp(sr)

matrix value of coefs, scores, variances, and t2 and

a histogram of "variance explained(%)" vs "principle component" will be displayed.

i would like to ask, based on this result, how should i know which principle is not contributed on the measurement, which i can then delete it out out of 20 principles? how can i relate PCA on my problem here.. i've read through alot of articles about PCA, but they just did not show out, how it actually can be applied in a simple problem like this.

hope i can get a clearer picture here.

thanks in advance!

[dspusa]

I have someone that will get back with some answers shortly. We are on break right now but will go over this when the students get back.

This will be a good test for 2 of our students.
Payday Loans
_________________
Domain Parking | Live prayer | RSS XML