Repository navigation
Analysis Module
The Analysis Module imports tracking, prepares cleaned tracking and metrics, optionally applies spatial or temporal filters, runs selected classifiers, and saves results beside the source data.
Review Prepare Inputs. Select the correct tracking type, enter a pixels-per-centimeter calibration, and choose the classifiers and smoothing options appropriate for the recording.

- Launch BehaviorDEPOTv1.7 from MATLAB's Apps tab.
- Select classifier rows to run. Use Edit only when you need to change classifier parameters from their defaults.
- Under Options, set the calibration, likelihood cutoff, smoothing method and span, and Hampel correction as needed.
- Optionally enable custom ROIs, cue-event filtering, and output plots.
- Choose Single session, then start analysis and select the tracking file and video.
- In the registration window, map each detected tracking label to a known body part. Named parts are unique; use Other for additional points. Review the prefilled mapping before confirming.
- Draw and name any requested ROIs. Complete every ROI before continuing.
- Review results in
<video-stem>_analyzedbeside the video.
The registration window displays an example tracking frame beside the mapping controls. Assign each detected label to the matching known part, then select Confirm Mapping. A named known part can be selected only once; select Other for an additional or unrecognized point.

When custom ROIs are enabled, draw and name each region in the integrated ROI window. Use Redraw ROI or Delete Current ROI before confirming if a region needs correction.

For multiple ROIs, the window retains previously completed regions while you draw and name the next one.

- Prepare the folder layout described in Inputs.
- Choose Batch process and start analysis.
- Select the batch root folder.
- Complete keypoint registration and ROI prompts for the first session.
- When prompted, choose whether later sessions reuse ROI names and limits, names only, or neither. Choose whether cue names should be reused.
- Review the output folder created inside each session folder.
When the first session's registration is complete, choose whether later batch sessions should reuse the same tracked-part mapping.

For custom ROIs, choose whether to reuse both ROI names and limits, reuse names only, or define new ROIs for later sessions.

- Custom ROIs report the intersection of selected behavior and named spatial regions.
- Cue files report behavior relative to frame-locked events.
- Spatial, spatiotemporal, and behavior-location plots are optional visual summaries; they do not alter classifier output.
When prompted after selecting cue filtering, enter a concise name for the time-locked event.

Use a single analysis to test tracking, calibration, mappings, and parameters. Use batch processing only after a representative single session produces the expected result. See Help & Troubleshooting for input and dialog problems.