Uncovering surveillance planes with BuzzFeed

From a list of points along a flight's path, how can you say "this looks like a surveillance plane?" And once you've found them, what do you do with the results?

logistic regression feature engineering publishing ethics classification

Summary

In 2016, Buzzfeed News published a piece exploring government surveillance planes, which they found by combining a federal plane database with flight tracking information. One year later they followed this investigation up with another piece that used machine learning to uncover potential surveillance planes that weren't listed in the registry.

In this chapter we will dig deep into the process of feature engineering, the process of creating new ways of describing your data that improve the performance of your machine learning algorithms. We'll also explore the difference between using yes/no predictions as compared to probabilities. Most exciting of all, we'll make maps of flight paths!

Notebooks, Assignments, and Walkthroughs

Feature engineering - BuzzFeed spy planes

Flight paths come as a series of individual measurements. How do you combine them together to create a profile of the way an individual plane typically flies?

Drawing flight paths on maps with cartopy

After all this analysis, it's time for a break! Take the suspicious flight paths and draw them on a map.

Finding surveillance planes using random forests

After the feature engineering has been completed, use a random forest classifier to detect spy planes.