Sometime between 2015-2016, I wrote a sale price probability estimator for at-auction vehicles. For this, I also wrote a nice database schema to track these vehicles. The client company's data was a pain to work with, and required a rigging of SOAP scrapers and liberal INSERTs to organize it. The design/mockup and initial concept was done by Claire Rutkoske and Jonathan Daun.
The tool was composed of several components:
- Database (MySQL)
- HTTP API (Node.js)
- SPPE client (JS)
- User interface (HTML/CSS)
Notes about the SPPE client
The HTML page is really a little client application that has some smarts. The API request merely returns a list of vehicles, and it is up to the client to render what it should look like. Instead of showing a histogram, it shows a kernel density estimation curve which instead of plotting points, instead plots probability functions (the kernel). The resulting curve is the summed probability distribution.
See an early development screenshot