Introduction¶
Why Learn to Write Programs?¶
Python programming is the most useful scientific tool that I’ve learned outside of general laboratory skills. Spreadsheet programs, such as MS Excel, are capable tools that serve a useful role for many scientists; however, there are a number of reasons that it can be advantageous to learn a programming language as a compliment to a spreadsheet-based software. First of all, simple programs can automate many of your repetitive data analysis tasks. In addition, a programming language can be helpful for specialized data analyses, and with your programming skills, you have the ability to create your own custom tools for the job. As a bonus, learning to program can be an enjoyable pastime that also provides you with a clearer understanding of computer terminology and processes.
Why Choose Python?¶
There are many, many programming languages, so an obvious question is “Why choose Python?”. First of all, Python was designed with readability as a primary goal. As such, Python code is easier to read than almost any other programming language. Python has a minimal syntax with very few special characters (brackets, semicolons, etc.), and it enforces some structural uniformity, specifically indentation, to improve code appearance. A downside of these structural rules is a small lack of flexibility that can take some getting used to; however, given some practice, these constructs will become second nature. This readability standard makes Python code extremely rapid to write as well. This means that you will spend less time processing your data and more time on interpretation and analyses.
In addition, Python comes with an interactive interpreter that allows you to test code snippets and track what is happening to your data. This means that you don’t have to write your entire program to see if things are working. You can check different segments of code and track the results without having to do all of the work in advance. A third-party advanced interpreter (IPython) is also available that takes the interactive coding experience to a new level.
One of the most important aspects of Python, from our point of view, is its large and active scientific user base. In my opinion, scientific users are drawn to Python by the advantages outlined above, and the fact that it is open source is important for future reproducibility (a fact that can sometimes be overlooked when you are using proprietary software). These users have contributed huge amounts of freely-available code , which can be used as a base for all of your scientific needs. See Appendix A for a partial listing of some of these third party add-on packages for various scientific disciplines.
Why Another Tutorial?¶
I started teaching myself Python in 2007, and at the time, I found a near infinite number of introductory tutorials, both on the web and as books made of paper. Unfortunately, most of these reference materials are geared towards computer scientists and professional programmers, so I spent a substantial amount of time determining how to apply these concepts to my data analysis problems. This meant that, compared with a program like Excel, I was not able to productively apply this knowledge for quite some time. Below is a graphical productivity versus time investment comparison of Excel and Python from my personal experience.
(Source code, png, hires.png, pdf)
I wrote this document as a way for new users to learn Python programming in a more efficient manner. After gaining some programming experience, I now realize that many of the tasks I wanted to accomplish did not require an exhaustive understanding of the language. Other tutorials proceed in step by step manner with substantial detail spent on each subtopic. This is a useful approach to learning but is also very time consuming. In this introduction, I will start with a very brief introduction to some of the key features, constructs, and terminology associated with Python, but most of the advanced topics will be presented in the context of real world examples. Each of these examples requires a slightly different skill set, so new concepts, terminology, and third party packages will be presented as needed to get you up and running as quickly as possible. My hope is that that this modified pedagogical approach will improve your learning curve, making it more along the lines of using a spreadsheet-based program (see below). Unfortunately, using a programming language will never be as ‘easy’ as a spreadsheet, but you’ll see that the analysis power gained using programming techniques will far outweigh the time costs.
(Source code, png, hires.png, pdf)
Practice Makes Perfect¶
Duh. You will not learn anything by passively reading this material. It is essential that you actually work through these examples. In addition, find a project that you would like to tackle for your own research. Forcing yourself to apply this to something that is important to you will really help to solidify this material in your mind.
External Resources¶
This tutorial is not meant to be a comprehensive. Several reference sites are highlighted here that are more generally useful than this short document will ever be.
The official Python documentation: This is a great resource that you should get to know well. It has a nice, short tutorial to the language, and a reference for pretty much everything else.
Scipy Lecture Notes: This may be one of the best tutorials for learning Python in a scientific context. These are actively developed reference materials and are updated on a somewhat regular basis. I don’t like some of the organization of the material, but it is a good general reference for many of the third party scientific packages that are available for Python.
Stackoverflow: This is a general programming question/answer site; however, it gets very heavy scientific Python traffic. In addition, users can vote on the answers they like best, making it pretty easy to find the most useful information.
Google: Sometimes you just gotta Google that thing. Use this frequently. Before you seek out advice from a peer or mailing list, type your question into Google verbatim. You might be surprised what you find.