Multi-Agent Programming Contest 2016 - Result Charts
Charts from the Multi-Agent Programming Contest 2016 results.
They are online!
Each match between 2 teams resulted in 3 simulations of a 1000 steps each.
Each simulation record is stored in a folder with 1001 XML files referring to each simulation step.
Script preprocess.rb reads an entire simulation folder collecting information about team's money at each step to generate a single CSV file to be used as data for the charts.
This step may take time with a slow hard disk, but you can use the already preprocessed CSVs in the folders.
# 3 matches are inside the zip
unzip dayXmatchY.zip
# The script expects to be where you unzipped
ruby preprocess.rb matchX_folder
# CSV file is at matchX_folder/matchX_folder.csvThe CSV files look like this:
Step,BathTUB,Python-DTU
0,50000,50000
1,50000,50000
2,50000,50000
3,50000,50000
...
998,-640533,49725
999,-640533,49725
1000,-640533,49725
The CSV files can be used with Google Sheets to generate and publish the charts using Area charts (using row 1 as headers and column A as labels). Currently the charts are generated using Google Charts without Google Sheets. Although they look the same, Google charts use the files from the repository and generates the charts on the client-side instead of transferring charts from Google Sheets.
Script analyze.rb reads datasets to solve the following questions:
- How much of the money in game each team have (total and percentage)?
- How many steps each team dominated the game (total and percentage)?
- The result was expected? The team that dominated most of the game won the match?
The script works in two modes:
# Read all CSV files in folders at current dir
ruby analyze.rb
# Read specified CSV files (one or more)
ruby analyze.rb "day1match1 BathTUB Python-DTU/City2016_BathTUBPython-DTU_2016-tournament-sim1of3_2016-09-12_13-00.csv"Both modes output results like this to screen:
____BathTUB vs Python-DTU - simulation 1 of 3_____
BathTUB made -640533$ (-92.80%)
Python-DTU made 49725$ (7.20%)
BathTUB dominated 0 steps (0.00%)
Python-DTU dominated 988 steps (98.80%)
draw dominated 12 steps (1.20%)
Expected result (won most steps)