Three-Box Foraging Behavioral Analysis

This is where I perform all my analyses of the behavioral data in the three-box foraging task. This repo is organized into the following sections:

1. data - this folder contains experiment data as well as any data structures generated by the code. Experiment data can be downloaded upon request and placed in data/experiments. Pre-existing data structures useful for analysis may be found under data/analysis.
2. docs - this folder contains useful documentation about the experiment, results, etc. Some of the files are directly generated from notebooks contained in notebooks.
3. figures - this folder is used to contain any figures generated by scripts.
4. notebooks - this folder contains jupyter notebooks that interface with the foraging package to render results and figures.
   • start with the Data Cleaning.ipynb notebook first to get an overview of the data and how it's cleaned. Then proceed to the Data Exploration and then the Learning notebook.
5. src - this folder contains the foraging package, which you should install locally in your environment. See steps to setup below.

Dependencies

You can see the full list in the pyproject.toml under the dependencies field. There are also optional dev dependencies listed under project.optional-dependencies.

Setup

Environment

Here are a couple tested ways to quickly set up an environment

1. Unidep

With Unidep, you can install conda + pip dependencies in one line. It can even install the foraging project using pyproject.toml. Just run:

unidep install -n <YOUR_ENV> pyproject.toml

or

unidep install -e -n <YOUR_ENV> pyproject.toml

to install in editable mode.

2. Conda

Create a conda environment from environment.yml by running:

conda env create --name <YOUR_ENV> --file environment.yml

Note you'll still need to install the foraging package locally.

Install foraging

• This step assumes you have built a functional environment and are ready to install the package
• Cd to the root of this repo and just run:

pip install .

• To install in editable mode, where you may wish to edit the package files, run:

pip install -e .

Tricks

Some useful tricks for this project are detailed here.

Generating HTML from notebook

After running the percent .py notebooks, you can use jupytext to convert them to whatever you like, including HTML. Personally, I like keeping a local .ipynb notebook that is paired with a percent.py notebook and converting the .ipynb to HTML without code by running

jupyter nbconvert --to html <notebook>.ipynb --TemplateExporter.exclude_input=True

#TODO: add task section and experiment section

Contact: [email protected]