Skip to main content

Outlier detection

Outlier detection estimates distribution of each variable and then identifies points in the data that have a low likelihood of belonging to the fitted distribution. Outlier detection is performed for each column individually, i.e. it is a univariate method.

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 consits of following steps:

The detect method consists of following steps:

  • get distribution fit from corresponding model
  • calculate anomaly indicator for given points
  • detect outliers

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