Post Classification Validation

Overview

VIDEO TUTORIAL: Random Forest Model

The results from running your model over a playlist can be further validated to eliminate any potential false positives. This is called a post-classification validation.

Tip: Random Forest Models run optimally on both Firefox and Chrome.

Get Started

To create a post-classification validation, first create a playlist with all the positive detections from running your model, depending on the used approach (RF, Threshold, Combined, or the three of them).

Listen to all or a subset of recordings to remove the false positive detections.

Next, run your model on this new playlist with the false-positive detections removed. Now you'll have results with fewer false positives.