I use RRCF to detect anomalies in a streaming 6-dimensional dataset where prices for 6 related products come in all the time. Since their distribution is not stationary but have a trend (due to inflation, the prices for products are going up) - would it be a good idea to scale the data to [0, 1] before applying the algorithm?
I use RRCF to detect anomalies in a streaming 6-dimensional dataset where prices for 6 related products come in all the time. Since their distribution is not stationary but have a trend (due to inflation, the prices for products are going up) - would it be a good idea to scale the data to [0, 1] before applying the algorithm?