Data Workshop Environment

What?

Basic environment for Data Workshop

Why?

Minimum effort & maximum impact :squirrel:

• Jupyter Notebook and Jupyter Lab
• Anaconda with Python 3.6.0
• Additional packages seaborn, ggplot, hyperopt, hyperas, ml_metrics, xgboost

Prerequisites

Only Docker (installation Instruction for Mac and Windows)

Build

docker run --net=host --dns 127.0.0.1 --dns 8.8.8.8 --dns 8.8.4.4 -dit -p 8888-8889:8888-8889 --name dataworkshop-environment dataworkshop/environment

Use

• Notebook - localhost:8888
• Lab - localhost:8889

Note: if you're a happy Docker Toolbox user to find the ip address use docker-machine ls.

Example

The URL column (docker-machine ls) contains tcp://192.168.99.100:2376, so you should copy 192.168.99.100 and add notebook port 192.168.99.100**:8888** or lab port 192.168.99.100**:8889**.

Re-use (already built container)

docker start dataworkshop-environment

Note: that in docker terminology

• run means build (a new container)
• start means start (already exists) container

Stop

docker stop dataworkshop-environment

Update (image)

To get the last changes from dockerhub

docker pull dataworkshop/environment

Remove container/image

container

docker rm dataworkshop-environment

or image

docker rmi dataworkshop/environment

Runtime metrics

docker stats dataworkshop-environment

Show Running Processes

docker top dataworkshop-environment

For geek

Note: run it in Dockerfile directory

docker build -t dataworkshop-environment .

Running container without jupyter server

Jupyter servers are started by running start_script.sh in CMD section. However, you can easily override it, by specifying a command at the end of docker run... (in this case servers won't start)