ml-docker

An all-in-one Docker image for machine learning.

Specs

This is what you get out of the box when you create a container with the provided image/Dockerfile:

• Ubuntu 18.04
• GCC/G++ 7
• Java 8
• Jupyter Notebook
• Conda Python 3.6.x
• Python Data Analysis : Numpy, Pandas, Matplotlib, SciPy, Seaborn, Yellowbrick
• Python Machine Learning : scikit-learn, xgboost, LightGBM, Catboost, mlxtend
• Python NLP : gensim, spaCy
• Python Auto-ML : TPOT, h2o-automl
• Python Hyper-Optimizer : hyperopt, sklearn-deap
• Python Deep Learning : Tensorflow, Keras
• Python interpretability : shap, skope-rules, lime
• Python Genetic Programming : gplearn
• Python Topological Data Analysis (TDA) : KeplerMapper
• h2o plateform : h2o

---

Setup

Prerequisites

Install Docker following the installation guide for your platform: here

---

Obtaining the Docker image

• Option 1: Download the Docker image from Docker Hub

Available here : (https://hub.docker.com/r/nielsborie/ml-docker/)

docker pull nielsborie/ml-docker

• Option 2: Build the Docker image locally

1. clone the repository

git clone https://github.com/nielsborie/ml-docker.git
cd /ml-docker

2. basic building :

docker build . -t ml-docker

3. If you have a proxy issue execute the following line :

docker build . --no-cache --force-rm --build-arg http_proxy=<proxy> --build-arg https_proxy=<proxy> --build-arg no_proxy=localhost,<proxy>,<proxy>,.an.local -t ml-docker

---

Running the Docker image as a Container

Once we've built the image, we have all the frameworks we need installed in it. We can now spin up one or more containers using this image.

Note:

• This images is based on jupyter-tensorflow docker.
• By default the image automatically launches a jupyter notebook on port 8887 of your localhost.

• Basic run

Simplest command to launch the container.

docker run --name ML-env -p 8887:8888 nielsborie/ml-docker

If you don't want to be stick to your terminal, you can run it in detached mode (-d)

docker run --name ML-env -d -p 8887:8888 nielsborie/ml-docker

• Start & Stop

Once you create the container, all you need to do is launch it :

docker start ML-env
docker stop ML-env

• Enter in the running container

Since you launched the container, you can get in and be root inside the image!

docker exec -it ML-env /bin/bash

• Customize your Container

• If you want a real password (and avoid copy/paste token step...)

docker run --name ML-env -d -p 8887:8888 -d nielsborie/ml-docker start-notebook.sh --NotebookApp.password="sha1:b6dba7097c97:7bded30fcbd5089adb3b63496d5e68921e102a5f" 

Note:

• default password = admin

• If you want to share your current working folder, you can map it with "-v" or "--volume"

docker run --name ML-env -p 8887:8888 -d -v /sharedfolder:/home/jovyan/work/ -e NB_UID=<your-UID> --user root nielsborie/ml-docker start-notebook.sh --NotebookApp.password="sha1:b6dba7097c97:7bded30fcbd5089adb3b63496d5e68921e102a5f"

Parameter	Explanation
-it	This creates an interactive terminal you can use to iteract with your container
--name	This set a name to our container, in our case we use ML-env but you can change it
-p 8887:8888	This exposes the ports inside the container so they can be accessed from the host. The format is -p <host-port>:<container-port>. The default jupyter notebook runs on port 8888
-v /sharedfolder:/root/sharedfolder/	This shares the folder /sharedfolder on your host machine to /home/jovyan/work/sharedfolder/ inside your container. Any data written to this folder by the container will be persistent. You can modify this to anything of the format -v /local/shared/folder:/shared/folder/in/container/
-e NB_UID=<your-UID> --user root	This fix permission issues under the container, you need to replace with your UID. You can get it with : id -u
nielsborie/ml-docker	This the image that you want to run. The format is image:tag. In our case, we use the image ml-docker and tag latest
start-notebook.sh --NotebookApp.password	It allows to launch the jupyter with a password already configured to admin

---

If you need to change the password check :

Jupyter notebook documentation

You can find additionnal info here :

Jupyter docker documentation

---