I am working on a big training dataset (21007 rows, 62 columns) and to reduce the dimensions, I\'m using PCA from the sklearn package. I\'ve imputed the null values (using K