Skip to main content

Kolmogorov-Smirnov

Two sample Kolmogorov-Smirnov test for drift detection. The detection is applied for each column individually.

Firstly the maximal distance between cumulative distribution functions (CDF) is calculated. For the given distance and sample sizes the p-value is calculated and compared with given p-value threshold.

For more details check:

  • Configuration: all the parameters that are available to be adjusted to the user's specific needs.
  • Outputs: understand TIM's outputs.

Engine schema​

The build-model method consists of following steps:

  • calculate CDFs for reference and test data
  • calculate maximal distance between CDFs
  • detect drift with respect to the p-value threshold

The detect method consists of following steps:

  • get reference data and p-value threshold from model
  • calculate CDF for reference and test data
  • calculate maximal distance between CDFs
  • detect drift with respect to the p-value threshold

Jobs of types rebuild-model, what-if, rca and upload model are currently not supported for jobs with drift approach.