OA: Hyperparameter tuning of stream algorithms

[detlefarend]

Description/Motivation
In benchmarking scenarios, a comparison of algorithms is only meaningful if they are optimally parameterized.

Brainstorming

• Benchmark stream generators could label their instances (e.g., point anomalies could be labelled with "true" or "1", ...)
• Scoring for anomaly detection: number of correctly identified anomalies, number of incorrectly identified anomalies, total number of anomalies as a reference

Task list

• 1. New class OAStreamTraining
• 2. New class OAStreamTrainingResults
• 3. Review/extension of classes OAStreamTask, OAStreamWorkflow regarding dealing with hyperparameters

Related issues
#...

Cross references
...