title
Coati ACC Pipeline- README

Coati_ACC_Pipeline

Predicts behavioural states from coati accelerometer data, among other things.

The following directory and file structure is needed for this program to run:

	PROJECTROOT/
	├── code/
	├── Data/
	│   ├── ethogram.tsv
	│   ├── acc/
	│   └── audits/
	├── Figures/
	└── tools/

The files in Data/acc must be generated using e-Obs DataDecoder, as specified below. The files in Data/audits must be generated using BORIS, and config.py must reflect appropriate behaviours in the lists STATES, EVENTS, and METALABELS.

Important notes:

1. Before using this software, use the DataDecoder provided by e-Obs to extract ACC data in columns with time data. To do this, first download the DataDecoder from here. (The linux version is already available in the tools/ directory here. Then, extract data from logger.bin files using the command

   decoder -f <filename> -d <output-dir> -c 6 -a c
   

2. This software uses the locale package in python, and sets it to de_DE.utf8. This change is necessary because the e-Obs DataDecoder saves time format with day-of-week in German (So, Mo, Di, Mi, Do, Fr, Sa).

   To do this on linux, you must first install the German locale

   sudo apt install language-pack-de