Work in progress
I wan't to build a statistical model to solve this problem, formally. I want to solve it since I want to draw nice pictures of airplanes. Everyone likes nice plots.
The likelihood will be pretty complicated, so I will use Stan for the parameter estimation.
But first, I need some airplane survival data.
Before you want to model data, try to simulate a similar data set. It will force you to think about the problem.
-- Brian Ripley
To simulate this, I need a way to draw little airplanes, and then I need to be able to shoot at them. Turns out I can do most of this using Affine_transformations on polygons. Making it into a little physics engine isn't that far removed.
The file test_flight.py draws the plane and a random flight path.
The file test_plane_and_shooting.py draws the plane and takes some random shots at it.
Next step is to work on a damage process with which to simulate killing the planes. Finally, I can try to infer that damage process using statistical modeling, to see where the plane's weak points are (and help Wald determine where to re-enforce the armour).
Will also give this a proper project structure later.
The file test_generate_destruction_data.py uses a damage model to shoot at and destroy planes. Some sample data is given in the directory plane_plots, as images and json files of the hits.
These can be analysed to figure out where the plane is weakest. Remember, Wald didn't get to see the destroyed planes, only the survivors.