The generation of observation results for the calculated feature is usually the most resource demanding process during the execution of calculated features. Therefore the Application Server tries to detect if changes (modify, add, delete) were made to observations of the source features since the last execution of the algorithm. If no changes are detected there is no need to execute the algorithm. In that case the success status in the execution history will still be success.

Instantiating calculated features will skip the detection of observation changes and will always execute the algorithm. I.e. even though no data was changed a new feature instance will be created and observations will be generated.