zizou 0.0.5

Python package to detect anomalies in geoscience time series data

Zizou: time-series classification tools for Geoscience

Zizou is a package to provide common tools for time-series classification in Geoscience. For example, it allows you to download seismic data from an S3 bucket or an FDSN server, compute spectrograms, and then use an Autoencoder to detect anomalies in the spectrograms.

Requirements

• xarray
• boto3
• pandas
• numpy
• scipy
• obspy
• tqdm
• xarray
• pyyaml
• tonik

Using the machine learning modules requires the following additional packages:

• scikit-learn
• pytorch

Installation

To only compute features run:

pip install -U zizou

To also use the machine learning modules run:

pip install -U "zizou[ML]"

Installation from source

Setup conda environment

cd zizou 
conda env create -f environment.yml

Install package in new environment

conda activate zizou 
cd zizou 
pip install -e .

Run tests

To run only the quick tests:

cd zizou 
pytest

To run the whole test suite:

cd zizou 
pytest --runslow

Setup Jupyter notebook kernel:

conda activate zizou 
python -m ipykernel install --user --name zizou 
kernda -o -y /path/to/jupyter/kernels/zizou/kernel.json

To find the path of your kernel.json file you can run:

jupyter --paths