Task Type, defined at the solution creation step when using dataset-based pipelines, can be Regression, Binary Classification, Multi-class Classification, or Anomaly Detection.
Select the appropriate Metric to assess model performance.
The default metric is Accuracy for classification, RMSE for regression.
See the supported task types, and the metrics available for each one:
| Task type | Description | Available metrics |
|---|---|---|
| Binary Classification | The model predicts one of two possible classes (for example, "yes/no", "healthy/unhealthy"). | Accuracy, AUC, Balanced Accuracy, F1, Gini, Lift, LogLoss, Precision, Recall |
| Multi-class Classification | The model predicts one of three or more possible categories (for example, types of activities, types of events). | Accuracy, Balanced Accuracy, F1 (weighted, macro), LogLoss, Precision (weighted, macro), Recall (weighted, macro) |
| Regression | The model predicts a continuous numerical value (for example, temperature, heart rate, pressure level). | MAE, MSE, R², RMSE, RMSLE, RMSPE |
| Anomaly Detection | The model identifies unusual patterns or outliers that do not conform to expected behavior (for example, detecting faults in machinery, gear shifts in sensor data). | Reconstruction Accuracy |