The goal of this project is to make a winning strategy for PLTW CSE Rock Paper Scissors. The network uses LSTM to do pattern detection and attempts to guess the opponent's next move.
I am no longer a CSE student, but I wanted to see if I could make a network that would destroy.
Last 10 moves (opponent's and network's) (3 booleans, one for rock, one for paper, one for scissors) Represented with a numpy array of shape (10,6) ((20,3) in the old versions, which doesn't work so well)
This network was trained on data from actual rounds of PLTW rps (from students) and on generated data makedata.py (from basics strats).
In the future, I might like to train the network with data from games that it has played, but I haven't done this yet.
Three outputs, one represents rock, one paper, and one scissors. This represents the networks prediction for the opponent's next move.
It works pretty well! tester.py uses the local implementation of the network in team.py and tests against other (basic) algorithms in opponents.py.
OVER ALL RESULTS
wins: 824
losses: 104
ties: 72
win rate: 0.824
More specific details
OPPONENT NAME : opponent_random
NETWORK : rspprsrsrssssprpsprpsrrrspprsssrprsrpsspprsrrrrsrrssrrrrrppsprsrrppsprsrprprspsrprpssprssrpssssssrrr
PREDICT : sprrspspspppprsrprsrpsssprrspppsrspsrpprrspsssspssppsssssrrprspssrrprspsrsrsprpsrsrpprsppsrppppppsss
OPPONENT: rpssrpsrrpsspssspsrrppssprrpspsrpprssrsssrssppsspsrprrpprrsppssrrsprrspsssrppspprrsrsrrrrrsppsprrrss
wins: 38
losses: 34
ties: 28
win rate: 0.38
OPPONENT NAME : opponent_beat_last
NETWORK : rsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprspr
PREDICT : sprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprs
OPPONENT: rprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprs
wins: 99
losses: 0
ties: 1
win rate: 0.99
OPPONENT NAME : opponent_repeat_rs
NETWORK : rsppprsrprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprpr
PREDICT : sprrrspsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrs
OPPONENT: rsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrs
wins: 96
losses: 2
ties: 2
win rate: 0.96
OPPONENT NAME : opponent_repeat_pr
NETWORK : rsrrsrprssspppssssssspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspsp
PREDICT : spsspsrsppprrrppppppprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprpr
OPPONENT: prprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprprpr
wins: 88
losses: 7
ties: 5
win rate: 0.88
OPPONENT NAME : opponent_repeat_sp
NETWORK : rrspprrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrsrs
PREDICT : ssprrsspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspsp
OPPONENT: spspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspspsp
wins: 95
losses: 3
ties: 2
win rate: 0.95
OPPONENT NAME : opponent_repeat_rps
NETWORK : rsppppssrssrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrp
PREDICT : sprrrrppsppsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsr
OPPONENT: rpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsr
wins: 94
losses: 4
ties: 2
win rate: 0.94
OPPONENT NAME : opponent_repeat_prs
NETWORK : rsrpsrspprsppprrprsprrsppprrprsprrsppprrprsprrsppprrprsprrsppprrprsprrsppprrprsprrsppprrprsprrsppprr
PREDICT : spsrpsprrsprrrssrsprssprrrssrsprssprrrssrsprssprrrssrsprssprrrssrsprssprrrssrsprssprrrssrsprssprrrss
OPPONENT: prsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsprsp
wins: 55
losses: 36
ties: 9
win rate: 0.55
OPPONENT NAME : opponent_repeat_srp
NETWORK : rrprssrpsrpprpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsr
PREDICT : ssrsppsrpsrrsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrps
OPPONENT: srpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrps
wins: 96
losses: 1
ties: 3
win rate: 0.96
OPPONENT NAME : opponent_copy_last
NETWORK : rrprprpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrp
PREDICT : ssrsrsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsr
OPPONENT: srrprprpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsrpsr
wins: 97
losses: 2
ties: 1
win rate: 0.97
OPPONENT NAME : opponent_antibeat_last
NETWORK : rrrrpsrprprrrsssppsrrpprrsspprrrrrssspssrprspprrrrsssppprrrrrssrrrrrrprrrrrrrrrrppppppssssssssssssss
PREDICT : ssssrpsrsrsssppprrpssrrsspprrsssssppprppsrsprrssssppprrrsssssppssssssrssssssssssrrrrrrpppppppppppppp
OPPONENT: sssssrpsrsrsssppprrpssrrsspprrsssssppprppsrsprrssssppprrrsssssppssssssrssssssssssrrrrrrppppppppppppp
wins: 66
losses: 15
ties: 19
win rate: 0.66
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent