Quantification of measured behavior parameters - this needs to be rewritten

[yavorska-iryna]

Before diving into pose estimation metrics, it would be wise to: a. QC our current behavioral measures (pupil, running, and licking) b. Perform descriptive analysis to visualize our these metrics change over time with learning and differ across animals.

In order to achieve this there are a few checks/steps that need to happen:

• [ ] Resample the traces for these three behavioral measures to be at the same sampling rate
• [ ] Remove outliers if needed (pupil and running traces have "spikes" that are an artifact of tracking tools, these can cause an issue during clustering of motifs)

These steps can follow:

• [ ] Create metadata dataframe to easily relate sets of motifs with training stages, mouse information, performance, etc.
• A few analysis tools to try:
• [ ] PCA
• [ ] Non-linear dimensionality reduction technique
• [ ] Others?

07.08.2024

• PRs were reviewed and merged (see below)
• code ocean capsule created with example of VBA behavioral data analysis and data plots. Some mice show distinct changes in running aligned to change stimulus, others dont. Here are two examples:

Running traces aligned to all images looks less informative. Main feature that I noticed is mice run a log more during Novel session. Some examples:

![Image]

(https://github.com/AllenInstitute/brain_observatory_analysis/assets/57956917/c5866e42-1f61-43bc-a4ff-69dba3c477b0)

[yavorska-iryna]

10.20.23

• start_lamf_analysis() to load new mice PR

[yavorska-iryna]

11.13.23

• Paused on this analysis. Vergil is currently looking at pupil and running dynamics.
• Requested reviews for PR that would load new mice.