README.md

Introduction to Python for the Macke lab

This repository collects some materials for visitors and new members of the lab to get up to speed on basic Python, with a particular focus on the type of machine learning applications we use here.

What you will find here

A series of Jupyter notebooks introducing Python and machine learning

• Python 101: Super basic introduction to Python.
• Data Analysis 101: An introduction to Numpy, Scipy and Matplotlib. If you have some basic programming experience, especially in Python, you should probably start here.
• Python 102: Some slightly more advanced techniques that crop up in virtually every real application. (In progress.)
• Regression - Orientation tuning: An introduction to least-squares fitting, using orientation tuning as an example.
• Machine Learning 101: [ TODO ] (Might be combined with Neural networks 101.)
• Neural Networks 101: A practical introduction to neural networks for classification. In this tutorial you build a 3-layer neural network, and experiment with different sizes for the hidden layer.

What you will need

Before getting started, you should install the following:

• Numpy
• Matplotlib
• Jupyter

Instructions for Linux

Installing with pip.

If this works, it's preferable.

pip install --user numpy
pip install --user scipy
pip install --user matplotlib
pip install --user jupyter
pip install --user scikit-learn

Installing from the package manager.

If you are on Ubuntu,

apt-get install python3-scipy
apt-get install python3-matplotlib  #(is this included in scipy?)
#apt-get install (Jupyter?)
pip install --user scikit-learn

Installing using Anaconda

It's also possible to install the Anaconda distributing, which includes everything you need (Python + all the packages mentioned above). The package versions in Anaconda tend to lag a little behind since they do their own packaging, but as long as you don't need the absolute newest versions, this is perfectly fine. Follow the instructions from here: https://docs.continuum.io/anaconda/install-linux.html.

Instructions for OS X

For the most pain-free installation, go with the Anaconda distribution. The installation instructions are here: https://docs.continuum.io/anaconda/install-macos.html.

You can also install through Homebrew, if you need the most recent versions.

Instructions for Windows

Compiling your own software is notoriously non-trivial on Windows, so unless you have a good reason not to, go with the Anaconda. Instructions are found here: https://docs.continuum.io/anaconda/install-windows.html.

After installing Anaconda

After installing Anaconda, you may want to install scikit-learn package to run some of the examples in the Data Analysis 101.

conda install scikit-learn

Acknowledgements

This set of materials was put together by relying heavily on similar works by

• Rick Muller
• Denny Britz

External references

Based on a list compiled by Rick Muller

Learning Resources

• Official Python Documentation, including
  • Python Tutorial
  • Python Language Reference
• IPython tutorial.
• Learn Python The Hard Way. If you want to invest the time to learn good all-around programming in Python, and not just writing quick data analysis scripts, this is one of the best resources.
• How to Think Like a Computer Scientist: Learning with Python A more classic text for learning Python. Make sure to read the Python 3 version.
• Python Module of the Week is a series going through in-depth analysis of the Python standard library in a very easy to understand way.
• Invent With Python, probably best for kids.
• Python Functional Programming HOWTO
• The Structure and Interpretation of Computer Programs, will deepen your understanding of general programming principles. There is no Python in this book (the examples are written in Scheme, a Lisp dialect).

Badass IPython Notebooks

• A gallery of interesting IPython Notebooks
• Each chapter of Rob Johansson's Numerical Python can be downloaded as a notebook. Note that these use Python 2.
• XKCD style graphs in matplotlib;

Packages for Scientists

Important libraries

• Python version 3.x (whatever 'x' currently is);
• Numpy, the core numerical extensions for linear algebra and multidimensional arrays;
• Scipy, additional libraries for scientific programming;
• Matplotlib, excellent plotting and graphing libraries;
• IPython, with the additional libraries required for the notebook interface.
• Sympy, symbolic math in Python
• Pandas library for big data in Python

Other packages of interest [ TODO Add ML packages here ] * * *

Cool Stuff

• Project Euler, programming problems that would (?) have interested Euler. Python is one of the most commonly used languages there.