Procedural Python for Reproducibility

The goal today is to take some of the ideas we developed last week, and do a couple things to make our lives easier:

• Define standalone functions which accomplish tasks that we would like to do repeatedly
• Put these functions in a place where we can easily use them without copy-pasting repeatedly
• Think about how to make our analysis reproducible – both for the sake of our future selves, and for the sake of anyone who wants to replicate and/or build on our work.

To that end, we are going to work together to do the following tasks:

1. Write a function which will download the Pronto data and the weather data

   • This function should only download the data if it isn't already present on disk
   • Pronto data can be downloaded at https://data.seattle.gov/api/views/tw7j-dfaw/rows.csv?accessType=DOWNLOAD
   • Weather data can be downloaded at http://uwseds.github.io/data/pronto_weather.csv

2. Write two functions which, given the downloaded data, will load it, parse dates properly, and return a pandas array.

3. Write a function which will group and join the trip and weather data into a single DataFrame, making use of the above functions.

4. Develop some plots showing relationships in the data, and write a function which will create and save plots related to your analysis.

   • Number of rides per day over the course of the year (day-pass and annual members)
   • Number of rides per hour over the course of the day (day-pass and annual members)
   • Number of rides per day as a function of temperature (day-pass and annual members)

5. Write a master script that you – or anyone – can run, which will produce your analysis from scratch.

Today during the class time we will walk through accomplishing these tasks together.