Project Score

Project Score Definition

The Project Score depicts the accuracy of the trained model for the project. Specifically, it is computed as the performance of the active model on the test subset. In a chained-task project, the Project Score is obtained by averaging the active models of each task.

Metric per Task Type

The following defines the metric that is used to evaluate the models per task type:

• Classification and Anomaly: Accuracy

• Detection: F-measure

• Segmentation: Dice Score

Usage Considerations

Given the variability in evaluation metrics, the Project Score should be interpreted as an indicative rather than an absolute measure. In addition to this quantitative method of measuring performance, the qualitative approach should also be considered, which consists of going through the predictions that the model produces. It is possible that the evaluation metrics are not showing a high number, while qualitatively, the predictions from the model are sufficient for the problem.