Feature selection using F-Test

5 次查看(过去 30 天)
I am trying to rank features given in following feature vectors, samples from three different class are given. I want to ask if it possible to rank them based on the F-Test score. I am new here and have gaps in understanding. I tried using anovan but I am not sure is it applicable here. I seek your guaidance and help with come simple example. I would appriciate is some example of feature ranking on IRIS dataset is possible.
class1 = [1.8779,2.651,2.2211,2.1688]';
class2 = [1.301,2.2902,2.3023,2.1251]';
class3 = [1.9504,2.4366,3.0451,2.9937]';
  2 个评论
Image Analyst
Image Analyst 2018-12-10
So you have some large data set, like an image or batch of measurements or something. Then you make 4 measurements of features on those samples, like feature 1 = mean value, feature 2 = median, feature 3 = average diameter, feature 4 = area fraction, or whatever they might really be.
So you have some vectors that define maybe the means of the measurements for class 1 and class 2. Or maybe those are sample vectors for collections of samples from class 1, and two collections of samples taken from what is known to be class 2 things. Or is the last one supposed to be class 3? Why is there no class 3 shown?
I think I need a bit more info before I can answer. How would you compute the F-test from each class vector that you showed?
Saifullah Khalid
Saifullah Khalid 2018-12-10
Sorry, it was a typing error, now I have corrected. I have samples from three classes. Each element of sample represent a feature like you wrote.

请先登录,再进行评论。

回答(0 个)

类别

Help CenterFile Exchange 中查找有关 Analysis of Variance and Covariance 的更多信息

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by