Over the summer I’ve been using Python for Machine Learning (ML), specifically state representation of an Atari 2600 game emulator. The project initially started with looking at this paper  which developed agents using Theano which then lead to me using the Anaconda Python package.
The package is great as it comes ready with everything most people would need to get started in Data Analysis and ML. As an example it comes ready with…
- Python 2.7 (3.x also available)
- Scikit Learn
- iPython Notebook
Once the project was over I started playing some of the ML functions including Clustering (K-Means) with Iris data and a simple Perceptron. Until this point I had been using an IDE (Spyder) and re-running the full batch of code everytime, however I discovered iPython Notebook which is so useful. You write the code in blocks and only run the blocks as needed. Fantastic for quick prototyping.
 Playing Atari Games with Deep Reinforcement Learning – http://www.cs.toronto.edu/~vmnih/docs/dqn.pdf