Run Model Over Playlist
VIDEO TUTORIAL: Random Forest Model
This is a classification method that uses the Random Forest model to classify the entire recordings dataset. The Random Forest model and Threshold model can both be applied to all recordings. Each model will classify the presence or absence of the species call in each recording.
Tip: Random Forest Models run optimally on both Firefox and Chrome.
1. Before running a Classification, create a playlist with the recordings that will be classified (e.g. nighttime, site or validated recordings, etc.). Each playlist should have no more than 20,000 recordings.
2. On the left menu, press Random Forest Models - Run Model and then click the + button to create a new classification.
3. Assign a classification name, select a model and a playlist.
4. Click Create, wait a few minutes and then click on the Refresh icon.
5. The status of each analysis can be viewed by clicking on Jobs.
6. The new classification will appear in the list, click on it and then click on Show Details to view the results.
Peaks indicate the similarity between the computed pattern (i.e. Classifier) and where the species is deemed present across the entire recording.
7. Click the Download icon to download results into a .csv file for the eventual use in statistical analyses.
You can opt to use the results of a Classification of just one approach (RF or Threshold) or a Combined approach, in which the positives detections that are common to both approaches are selected. The Random Forest approach usually provides the highest number of positive detections, but also a higher number of false positives than the other two approaches (Threshold and Combine approach). The Threshold and the Combined approach are more conservative approaches that usually provide a lower number of detections and false positives.