In response to an interview excercise, I created a project that showed my ability to produce multiple quality deliverables under a tight deadline.
Note
I used an API endpoint to retrieve the data, pandas to clean and munge, and bokeh to map inccidents. I also performed hypothesis testing and merged additional outside data. The final deliverable was an annotated slidedeck, a narrative data-processing notebook, and an interactive visualization of crashes.
The data for this effort is sourced from Chicago Data Portal Traffic Crashes.
Below is an approximation of the prompt given:
Imagine you are asked to create a short report for a Transit Authority management about the causes and trends among crashes in an urban enviroment, answering questions such as:
- When and where are crashes most likely?
- What types of crashes are most likely?
- Do you see any actionable insights for the management?
This report should be created assuming it will be presented to transportation leadership with little prior background of the dataset and it should not need additional explanation. You’ll be judged on the following criteria:
- How you processed the data (data cleaning and transformations).
- The design of the report (clarity and visual appeal).
- The quality of insight and the types of questions answered in the report.
Attaching any code or files used to create the report.