How can reduce features in dataset with existing algorithm in matlab

hi i have a data set for a cancer that has 68 features i want to create a model and reduce some of features some how the result don't change much how can i do that? it is possible with PCA ? what i should to do? please help me

Answer 1

[Walter Roberson]

http://www.mathworks.com/discovery/feature-selection.html

Answer 2

[Image Analyst]

The way I think of it is like each principal component is like a model - a weighted sum of your 68 original features. It could be that the majority of your cancer/normal prognosis is determined by just a few of those principal components, like PC1, PC2, and PC3. Then you can look at those and see what proportion of the 68 values goes into those. If you find that some feature contributes just a very small proportion to any of the PCs, then you might be able to ignore it since it's not a very predictive thing to measure.