This is a side project where I generate random data and cluster them according to their position in a graph.
This project was inspired by the idea of a 4X game (something like Sid Meier's Civilization V game) having an AI that attacked an opposing faction (now denoted as F) based on their distribution of units. Assuming the AI knew the positions of each of F's units (which is rather unlikely), the AI would determine the front that each unit was on. In this case, it could use the k-means clustering algorithm, setting k as the number of factions F was at war against. Then it would determine if F's units were in an optimal position and distribution for the AI to attack F.
However, this is not an optimal player experience as we assume that the AI knows the position of every unit of F, which is generally frowned upon in 4X gaming communities. We can remedy this by having the AI save the positions of each unit it sees of F and generating an uncertainty plot for each of those groupings of units. This uncertainty plot would change with each turn, as the AI would determine the possible new locations for F's units based on their known movement speeds and going in every direction that is possible. Of course, this uncertainty plot could very easily mean nothing after a small number of turns, but this would theoretically solve the problem of a dissatisfied fanbase as we no longer assume the AI knows all of the unit positions of F.
The project was built using packages installed from
Anaconda 4.5.11. The
packages that were used can be found in requirements.txt.