Greedy and EpsilonGreedy strategies, using Multi-armed Bandits algorithms#1438

Open

bing-j wants to merge 13 commits intoAxelrod-Python:devfrom

bing-j:epsilon-greedy-strategy

Commits on Mar 16, 2024

added and implemented epsilon_greedy.py
bing-j
committed
added and implemented epsilon_greedy.py
bing-j
committed
added and implemented epsilon_greedy.py
bing-j
committed
epsilon-greedy added to _strategies.py and reference\strategy_index.rst
bing-j
committed
Fixed update_rewards() to correctly refer to attributes, added test file
bing-j
committed
random test created and passed, basic deterministic tests created and passed
bing-j
committed
strategy tests completed and passedl; updated docstring to clarify initial behaviour.
bing-j
committed
References correctly added and docstring updated with correct Name.
bing-j
committed
for extendability: renamed module to armed_bandits, created Greedy superclass of EpsilonGreedy, and implemented option to weigh reward based on recency
bing-j
committed
modified UNIFORM to use np.inf instead; fixed EpsilonGreedy.strategy() so that it correctly calls update_rewards() through the parent method; all previous tests passed.
bing-j
committed
formatted armed_bandits.py using black.
bing-j
committed

Commits on Apr 22, 2024