I've reviewed these resources and highly recommend them. I periodically update this page, but it may be out of date.
Learning to use Python
If you want to learn to use python, here are the key skills that you need to know:
- Basic Python syntax, including the role of whitespace, variables, objects, functions and modules.
- Intermediate Python syntax, including lambdas, list comprehensions, generators, and decorators.
- Building unit tests with unittest or (better) pytest
- Python package management with setuptools and pip.
- Editing python programs with a decent editor, such as pycharm, or an interactive IDE such as spyder or Jupyter notebook. You can also use emacs or vim. Don't use Idle or Notepad++ or nano.
Do not learn python2.
To learn python you must be able to run Python programs. You have two main options:
- Although Python is pre-installed on many computers, do not use the pre-installed version, as it is typically out of date. (It's also frequently python2, which is abandoned.) You can instead download and install the Anaconda Python distribution. It has all of the packages you need.
- Alternatively, you can learn it entirely using web-based Jupyter notebooks.
If you don't know how to program
If you have never programmed with Python before, you may want to start with DataCamp's Learn python by example:
I also recommend:
If you know how to program
If you know another programming language (e.g. C, C++, FORTRAN, Java, etc.), you can learn Python in about 2 hours.
Start by reading the Python Tutorial:
I then recommend reading the documentation:
- Official Python Documentation. Just read the documentation. Honestly. It's well-written.
- Python Tips (book, online for free)
- Python Google Style Guide.
Recommended packages and tools for different tasks
- PyCharm, free for educational use (if you have a .edu email address)
- Jupyter Notebook (comes with Anaconda; also available https://jupyter.org/ on the web!)
Nice blog entries:
- Building a Financial Model with Pandas, looks at mortgage calculator
- Python for Finance: Analyzing Big Financial Data (O'Reilly)
- Financial Modelling in Python (Wiley Finance)
Numerical Data Processing
- numpy and scipy are the standard numerical analysis tools in python.
- pandas is a popular data analysis platform that integrates with numpy. You can make it run on a cluster with Dask.
- Alternatively, you can use Apache Spark and SparkSQL
- tabulate - create tables easily (but see my tytable replacement)
For working with dates, see:
- Use Pillow, the Python2/3 fork of Python Image Library (PIL).
- matplotlib - A well-developed system for plotting 2d and 3d static graphis. (See my Short example of using matplotlib within jupyter)
- holoview - a more modern but less featured system.
- Altair: Declarative Visualization in Python
- geopy - geocoding for Python
- Seaborn. (See Datacamp Tutorial)
- Graphics.py: http://mcsp.wartburg.edu/zelle/python/graphics.py http://mcsp.wartburg.edu/zelle/python/ http://mcsp.wartburg.edu/zelle/python/graphics/graphics.pdf
- HTML graphics: http://anh.cs.luc.edu/python/hands-on/3.1/handsonHtml/graphics.html
Python contains built-in support for SQLite3.
For using MySQL, you'll need a connector. There are many available.
Install MySQL connector on anaconda:
conda install mysql-connector-python
Then you can:
import mysql.connector as mysql c = mysql.connect(host=host,database=db,user=user,password=password)
I generally prefer PyMySQL, as it's pure-python and has fewer dependencies.
Python GUI Options
I recommend writing GUI's using a web browser as a front end and a local web server (see next section).
If you must write a native app, I recommend using:
- pyqt - Use PyQT5 (it's part of Anaconda)
- FLTK, the Fast Light Toolkit
However, I strongly recommend writing your python program as a web-based app (see next section)
Building Web Applications
You should use a framework that implements Python's Web Server Gateway Interface (WSGI) to write web apps. Popular implementations are:
- Bottle (easy, and the whole thing runs from a single file)
- Flask (can run on Apache with Passenger, but needs its own domain name)
Bottle has an built-in web server for development. For high performance you can combine it with CherryPy or use the Apache wsgi module.
For more information, see:
- See also https://wiki.python.org/moin/WebFrameworks
If you are using anything other CGI, you need some way to tell the web server to reload your python program. (You don't need to do this with CGI---it reloads Python and your program every time you serve another request. That's why it's so slow...)
Running WSGI on Dreamhost
To reload your program on Dreamhost with passenger:
$ touch tmp/restart.txt # reload your program
To reload your program with mod_wsgi:
$ touch wsgi_app.py
Using Passengers on Dreamhost:
Writing a full application with Apache, Bottle and MongoDB:
- Fluent Python: Clear, Concise, and Effective Programming (you can get it on Safari)