I have to find correlation and reduce my data, and remove the correlated value. Please anyone tell me how do i reduce my data set. I have weka installed in my system but i dont konw how it work, and i did not kno how to use it for finding correlation.
You should learn something about principal components analysis (PCA). You can then use the first few principal components instead of the 702 original variables.
Or you calculate all pair-wise correlations and rank them. You can then decide if you will take one variable of the pairs with highest correlation, or is you substitut the two variables by a new one calculated from the mean (for instance) of the pair-wise values (this will then be a somewhat similar approach as the PCA).
In any case you should get a rough idea if the relationships between the pais are linear. If not, you probably may want to transform some of the variables.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools