Simple Python 3 Machine Learning for Flash Game
New here? Learn about Bountify and follow @bountify to get notified of new bounties! x

Hi there!

Python 3 Machine Learning bounty.

So this bounty should be simple enough to solve for anyone with experience with machine learning, but the real focus is to create a program that not just works flawlessly, but importantly, the code can also be understood by the most inexerpienced coder possible.

The 'Club Penguin Rewritten' online flash game contains two minigames of interest, one called 'Puffle Roundup' and one called Card-Jitzsu.

Puffle Roundup:
This game requires the player to 'round up' the animals by placing the mouse behind them to push them in the direction of the open pen. There are multiple colours of puffles and each one has a different sensitivity (distance from cursor at which they will start moving). Puffles can be lost by being pushed outside of the boundary. Each time the next round is loaded, the puffles are distributed effectively at random.
The in-game reward is a flat reward per puffle sucessfully herded, plus additional points for the remaining time, so the program should be able to complete the task quickly. To avoid bagging just the nearest puffle and quiting, and to make the program practical, it should run rounds indefinitely until the program is terminated, thus being able to click the necessary button to begin the next round and recognise when to begin herding, when it is in a loading screen, etc, with the purpose of the program being to run continuously while optimising the number of points per time.
As mentioned at the top, it is essential the code is understood by a beginner programmer with only the most basic possible understanding of machine learning in Python 3. This means a good submssion will favour clarity over simplicity and contain comprehensive notes.

Create a program with the same expectation of being beginner-understandabe and Python 3 machine learning, to win at Card-Jitszu with better-than-random results, against real opponents. This program should be able to work given knowledge of any given set/collection of possible cards, which are either first read directly from the game or manually imputted. Again, this program should be able to repeatedly initiate and play games continuously until the program is terminate, given the player starts in the Dojo, then the program should click the Sensei's stand, click 'Earn your belts' and then play the game to completion.

Good luck!

awarded to KieranL

Crowdsource coding tasks.

1 Solution

Winning solution

Try Google opensource ML, they have a specific setup just for games