A PCA model is fit on a dataset with 10 features. The cumulative explained variance reaches 95% at 4 components. What does this mean?