Robust Scale

Objective

Robust scaling is less sensitive to outliers compared to standardization. It uses the median and interquartile range (IQR) instead of the mean and standard deviation.

Example

time window:
  emitting type: slidingRobustNormalization
  window functions:
    normalization:
      function:
        robust scale: {}
      attributes: [ ask,bid ]
  policy:
    type: slidingTime
    slide: 500
    window size: 2500

Output attributes

After processing, attributes receive automatically applied postfix ID = ROBUST_SCALE

Before processing	After processing	Type
ask	ask_LOG_NORM	double value
bid	bid_LOG_NORM	double value