Introduction

This repository exists to try out algorithms to build anomaly detectors using PCA. The repository consists of a Dockefile that would help launch a jupyter notebook. The jupyter notebook will be used to run notebooks to try out anomaly detectors that use PCA.

Run

Build the docker image to run Jupyter notebook.

docker-compose build jupyter

The above command will end up creating a docker image with the tag: ubuntu-18.04:jupyter

Run the jupyter notebook:

docker-compose up jupyter

Find the access token URI of the jupyter notebook session:

docker-compose exec jupyter jupyter notebook list

Connect to the jupyter notebook using the following URL:

http://localhost:9145/?token=<token>

NOTE: The docker container already maps the container port 9145 to the same port accessible through localhost. Hence, whatever URI you obtain from running the command jupyter notebook list you will replace the domain name of the URI with localhost to make the URI accessible.

Notebooks

Currently there is a single notebook in this repository learn-pca-fft.ipynb which is mounted as a volume in the container and can be directly accessed from the browser when accessing jupyter running in the container.