I have classified some data points as anomaly in timeseries data and calculated feature importance vectors for them. Now I want to calculate similarity between two anomalous data points based on their feature importance vector.