-
Notifications
You must be signed in to change notification settings - Fork 0
EvaluationResearch/ShapleyCompetitive
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
1.generatedata.py and generatedata4.py are for generating synthetic data, the worth of player are generated from a uniform distribution. generatedata4.py is for 4 players, and generatedata.py is for 5 players You can generate one test for each run, the data will be stored in Excel_test.xls, and you should copy them in 16or1.xlsx, test1.xlsx, and 16or2.xlsx,test2.xlsx 2. 16or1.xlsx, test1.xlsx, and 16or2.xlsx,test2.xlsx is the synthetic data generated by step 1 3. We have tested 3 sample methods: Random sampling, Stratified sampling,information-driven sampling For 4 players, we have shapleyForR.py,shapleyForS.py,shapleyForI.py for Random sampling, Stratified sampling,information-driven sampling For 5 players, we have shapleyForR5.py,shapleyForS5.py,shapleyForI5.py for Random sampling, Stratified sampling,information-driven sampling 4. R1, S1, I1 is the experiment result by 4 players, R51, S51, I51 is the experiment result by 5 player You can run splotF3.py to get the Spearman correlation and MSE between appropriate and real Shapley value with 3 methods. And splotFSM.py to get the Spearman correlation and MSE between appropriate and real Shapley value with each data get by step 3 5. SampleFormaddpg.py for sampling the coalition 6. We record all the results from MPE, and calculate the final result by shapleyFMPE.py 7. Plotf3.py is the approximating Shapley for predator-prey environment with increasing sample size.
About
No description, website, or topics provided.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published