Hi 晓兵 汤,
One way to think about it (call it 'naive PCA')
is to Rank your 'factors' by a meaningful Measure.
It could be as simple as the Sum of Differences (% points)
of the respective factors impact, giving the first Year as a baseline.
Please find the attached with just that.
Hope it could help.
PS One would weight by a Denominator (Sum of values in A) later.
Factor_Impact_v10.2_YF.twbx 48.0 KB