My primary software is Python, but I also code in R and SAS. For why Python continues to be my primary software, you can view slides and arguments from my presentation at 43rd Annual Conference of the International Society for Clinical Biostatistics at this LINK.

Despite Python being a powerhouse in data science, I’ve found less adoption among epidemiologists and biostatisticians. In case you wanted to get started with Python, the following is the setup I use on all my computers. This guide should get you started with coding in Python

Installation

While you can install Python from python.org, there is a better way. Here, we will use pyenv, which is a manager that allows us to easily use multiple versions of Python (I regularly switch between 3.6, 3.7, 3.8, 3.9, 3.10).

For detailed instructions on how to install pyenv you can view the guide for Mac/Linux at RealPython or the pyenv-win documentation for installation on Windows. I’ve used pyenv on both Windows and Linux.

Once pyenv is installed, we can install a specific version of Python. Let’s say we want to install v3.9.5. To do that, we would open terminal (Mac,Linux) or command prompt (Windows) and use the following command

pyenv install 3.9.5

This will take a little while, but you should see a message that says the install was completed.

After installation, you should be able to type the command python and have it bring up something like

Python 3.9.5 (tags/v3.9.5:0a7dcbd, ...)
Type "help", "copyright", "credits", or ...
>>>

Then we can type commands like

>>> print("hello world!")
hello world!
>>> x = 5
>>> x*10
50
>>> quit()

where quit() will exit Python.

Other versions of Python can be installed. To switch the python keyword between versions, see the other documentation at the RealPython link provided above.

Install Packages

Next, we are going to install the usual set of packages for data science. This set of packages contains the essentials to get started. We will be using the Python Package Index (PyPI) and pip to install packages. We are going to install the following packages

• numpy: numerical python for various calculations on arrays
• scipy: scientific python containing basic statistics, optimizers, and root-finders
• pandas: data management library
• matplotlib: visualizations
• statsmodels: statistical modeling with R-style formulas

To install a single package, we run

python -m pip install numpy

To install multiple packages at once

python -m pip install --upgrade pip

To update pip, we can run

python -m pip install numpy

Now Python is all setup for the basic tasks of data science. Various other packages can now be installed (including mine).

Warning: before installing any package from the internet (including PyPI), you should check the corresponding PyPI page and wherever the open-source code is made available. Only download packages you can trust.

Using Python

Now we are ready to start using Python. I have an introductory guide that works up to running statistical analyses and generating figures at this LINK. If you have any questions, feel free to open a GitHub issue on that page.