Table of Contents

Enum MetricOptimizationDirection

Namespace
AiDotNet.Enums
Assembly
AiDotNet.dll

Specifies the direction for metric optimization (whether lower or higher values are better).

public enum MetricOptimizationDirection

Fields

Maximize = 1

Higher metric values are better (e.g., accuracy, F1 score, AUC).

For Beginners: Use this when tracking metrics where larger numbers indicate better performance. For example, if accuracy goes from 85% to 90%, that's an improvement.

Minimize = 0

Lower metric values are better (e.g., loss, error rate, MSE).

For Beginners: Use this when tracking metrics where smaller numbers indicate better performance. For example, if you're tracking loss and it goes from 0.5 to 0.3, that's an improvement.

Remarks

For Beginners: When tracking metrics during training, you need to specify whether you want to minimize the metric (lower is better, like loss) or maximize it (higher is better, like accuracy). This enum lets you tell the system which direction represents improvement.

Examples:

  • Loss functions: Use Minimize (you want loss to go DOWN)
  • Accuracy: Use Maximize (you want accuracy to go UP)
  • Error rate: Use Minimize (you want errors to go DOWN)
  • F1 score: Use Maximize (you want F1 to go UP)
Edit this page