slack0 / presentation

Presentation for Kaizen Data Conference September 16, 2016

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

How I Use Docker for Data Science

Developing Models for Deployment

Prerequisites

Ensure that you have a working version of docker on your machine. Installation instructions for all major platforms can be found on the docker website.

You should be able to execute docker --version and see something like Docker version 1.10.3, build 20f81dd

If you are on linux, installation is very straight forward as docker takes advantage of the host's native kernel

If you are on Mac or Windows, you will need to use the docker-machine tools provided in the installation to run docker on a virtual machine. This is somewhat seamless, just follow the tutorials

Play with it

Take a look at the Hello World tutorial on docker.com

See what docker ps, docker logs, and docker stop do...

Development

You can view the presentation in Presentation.ipynb

We'll be using the repos:

Don't worry too much about what is in these repos just yet, but some familiarity will help.

Usage

If docker is properly installed, the workflow below will work

git clone https://github.com/mconley-kaizen/dockerapp.git
cd dockerapp
docker build -t mydockerapp .
cd ..
git clone https://github.com/mconley-kaizen/deploy.git
cd deploy
source launch.sh
deploy 5000 iris_prediction predict mydockerapp

If executed correctly, there should be a webapp running on port 5000 of your docker network. If you are on linux, your docker network will be your host computer. If you are using docker-machine, the command docker-machine ip will return your docker network, ie: chrome $(docker-machine ip):5000

About

Presentation for Kaizen Data Conference September 16, 2016


Languages

Language:HTML 97.7%Language:Jupyter Notebook 2.3%