Base Docker container to perform data analysis using:
- SQL (able to run locally using SQLite)
- Python
- accelaration using LLVM
- R
- Spark (able to run a local spark session)
The container is based on rpy2's container: it can run standalone on a single machine, making it well-suited for teaching, learning, or experimenting with the jupyter notebook work off-the-shelf.
Minimal examples of jupyter notebooks are in the directory notebooks. The container is used in workshops about polyglot approaches to the analysis of data (https://github.com/lgautier/odsc-ppda-slides or https://github.com/lgautier/jpd-pdapr-slides), or as an off-the-shelf solution to reproduce an analysis of data in a jupyter notebook (https://github.com/lgautier/project-tycho-utilities/blob/master/notebook/measles_and_diphteria.ipynb).
To have a Jupyter notebook on port 8888:
To start a Jupyter notebook:
docker run -p 8888:8888 --rm lgautier/pragmatic-polyglot-data-analysis
- Point a web browser to the URL
When running the Docker command a jupyter notebook will be started in the container. At the time of writing a token-based security setting is activated by default and the URL to use will be found in the output printed in the terminal (it will look like http://localhost:8888?token=<sequence of digits>).
Note: Our docker container is running with a non-root user jupyteruser.
Persistance of the work is highly desirable in many situations, and we can achieve it by mapping the host's user and group IDs with the IDs in the container. The following will start a jupyter notebook server that has access to the current working directory on the host. Note: check the jupyter notebook and docker security models, as well as where you are running your notebook (for example private or shared machine, private or public network) before deciding on doing this or choosing to use a particular working directory.
docker run -p 8888:8888 \
--rm \
-u $(id -u):$(id -g) \
-v `pwd`:/home/jupyteruser/work \
lgautier/pragmatic-polyglot-data-analysis