ML Lab

This reposotory contains some of my projects related to machine learning & data science.

Quick Start

Environment

I usually work with docker containers, so...

• Install Docker

• Clone this repository

$ git clone https://github.com/feliux/data_science_projects.git
$ cd notebook

Spark

Notebooks runs over jupyter/all-spark-notebook image.

• Pull the image from Docker Hub

$ docker pull jupyter/all-spark-notebook

• Run your container

$ docker run -d --name <container_name> -v $PWD:/home/jovyan/work -p 8888:8888 <your_name>/all-spark-notebook:latest

Or execute

$ docker-compose -f docker-compose-spark.yml up -d

Pytorch

You can use the previous container for this notebooks working with CPU. If you want to use your GPU just pull or build a new image with the Dockerfile I provide (based on the nvidia/cuda:10.2-base-ubuntu18.04 image). Feel free to add/remove whatever you want ;P

• Pull from Docker Hub or build the image

$ docker pull feliux/gpu-scipy-pytorch-notebook
OR
$ docker build --no-cache=true --build-arg BUILD_DATE=$(date -u +'%Y-%m-%dT%H:%M:%SZ') -t feliux/gpu-scipy-pytorch-notebook:1.0.0 Pytorch_Dockerfile/.

• Run your container. For example

$ docker run -d --gpus all --name gpu-pytorch -v $PWD:/home/jovyan/work -p 8888:8888 feliux/gpu-scipy-pytorch-notebook:1.0.0

Or execute

$ docker-compose -f docker-compose-pytorch.yml up -d

• Test installed packages (some examples)

$ docker exec -it gpu-pytorch python
>>> import torch; torch.cuda.is_available()
True

• Connect to your notebook

$ docker exec -t <container_name> jupyter-notebook list

• Paste the url/token in your favourite browser

References

Introduction to Hidden Markov Models with Python Networkx and Sklearn

Hidden Markov Model

Training the Perceptron with Scikit-Learn and TensorFlow

Graph Neural Networks

Python Speed

CNN

CNN