Overview of Project 2: Exploration & Path Diversity

[changsiyao]

Exploration & Path Diversity

What's the different moving patterns for each mouse strain?

[jarrodmillman]

For Monday, your task is to produce a 1 page summary here: https://github.com/berkeley-stat222/mousestyles/blob/master/doc/problem/path.md

Your goal is to understand the problem. You are welcome to look at the code, but you shouldn't be working with code at this point. You also probably don't want to focus on the algorithms yet either.

If this isn't clear, please reread the project assignment: https://github.com/berkeley-stat222/mousestyles/blob/master/doc/assignment/mouse.pdf

Feel free to discuss the issue on Piazza.

[changsiyao]

@pttzty @MengfeiJiang

[peterasujan]

Comments from class:

• We'll need to chunk the observations into a grid. What would be an appropriate grid size (the authors use 1cm)?
  • First, we need to decide what our goal is. One example is classification, which gives us an obvious measure of goodness (i.e. accuracy).
  • Can plot and decide if things look too chunky or too jittery.
  • Can see if temporal patterns emerge.
• Other considerations:
  • anomaly detection
  • may be able to detect different types of events using different grid sizes
  • directed actions vs. noisier actions
  • extract "vectors" of movement, which can include a similarity metric for different paths

[jcyhong]

See #35 for the structure of the summary.