/datascience-lab

Docker containers for Datascience

Primary LanguageJupyter Notebook

Docker for DataScientists.

My docker container for datascience.

DataScience docker

Main feature

  • Base docker image is deepo
  • Set Jupyterlab and install jupyterlab extensions.
  • Install orca docker container to fully use data visualization package plotly.

QuickStart

build docker image

sudo docker build --tag datalama/deepo:lab .

docker run

sudo docker run -itd -p 8888:8888 -p 6006-6015:6006-6015 --name lab --ipc=host datalama/deepo:lab jupyter lab --no-browser --ip=0.0.0.0 --allow-root --NotebookApp.token= --notebook-dir='/root'
  • I recommend add your volumes.

docker exec

docker exec -it lab /bin/zsh

Jupyterlab extension list

here's my favorites.

  • jupyterlab with ipywidgets packages(e.g. tqdm.notebook, plotly)
    • jupyter-widgets/jupyterlab-manager
  • To visualize plotly on jupyterlab
    • plotlywidget, jupyterlab-plotly
  • draw.io
    • jupyterlab-drawio
  • R studio style variable inspector
    • @lckr/jupyterlab_variableinspector
  • table of contents for jupyter lab
    • @jupyterlab/toc
  • jupyter lab extension for git
    • @jupyterlab/git

Install Jupyterlab extensions by hands

Sometimes, it is hard to build docker images with installing jupyterlab extensions. Then, I recommend install jupyterlab by hands.

  • Enable Extension Manager

  • Click Extension Manager(puzzle icon)

  • Search, Install and Build !!!

plotly orca setup

If you use plotly to EDA or visualize your data, sometimes you may want to export your figure. At that time, you need the orca(Electron app). Howerver, running orca in your container is a little bit tricky(chromium issue).

I recommend referring to this notebook.

NLP docker

QuickStart

docker run -itd -p 8888:8888  --name nlp-lab --ipc=host  datalama/lab:v2 jupyter lab --no-browser --ip=0.0.0.0 --allow-root --NotebookApp.token= --notebook-dir='/root'
  • jupyter 관련해서 좀 더 잘 사용하는 법을 알아보자