I am curerntly running PCA for returns series of 50 stocks for 524 observation. I have completed following steps like I have computed covariance matrix, got the loading for each of them and also respective scores for all 50 stocks. There are 20 component explaining 80% variation.
- When we get final scores for example
pc1 = -0.20*v1-0.50*v2.....-0.60*v50in this case we will multiply this weight to return series of this variable. If yes, then what would be final output?
- Is it fine to have all values negative in your first PC1?
- if someone can help me with R example to preceed further from Q1 it would really a great help.
- When my output says there are 20 component explaining 80% variation that means I need to derive
pc1+pc2....+pc20and the resultant value would be my final value is that correct ?