Darwinium Rules Sets and Models both produce scores in the range [-1000.00 .. +1000.00].
Scores above zero are indicative of "good" behaviour or Trust.
Scores below zero are indicative of "bad" behaviour or Risk.
The further the score is away from zero, the more Trust Worthy or Risky they become.
The Scores are classified into bands as follows:
Model Scoring
Machine Learning Models produce a probability score in the range [0.0 .. 1.0] with a specific point in this range (known as the Threshold) being the Threshold between desired and undesired behaviour.
Darwinium Models use 0.0 to indicate Risk and 1.0 to indicate Trust.
A scaling function is provided to scale the Probability output from the model into a Score relative to the Threshold.
For example, the diagram below illustrates the scaling of Probability based on a Threshold of 0.5 left and also a Threshold of 0.3 on the right:
Percentiles and Distribution
