🌱 Naught Robot's Digital Garden

# Ranking Board Games Based on Empirical Data

Category: Evergreen
Tags: Tabletop Games, Programing, Data Analysis

## š² Personal Game Ratings #

I love top ten list and board games. I’m also lazy and lacked a way to easily generate a top ten list of my favorite board games. Top ten list are naturally subjective to the individual and I’m looking for a little more empirical evidence as to why my number seven game is higher on the list then my number eight game. To accomplish this, I’ve combined my game ratings from Board Game Geek with the number of logged plays I have for each game to devise the most accurate top ten board game list for me. The assumption here is that if I really like a game, then Iām going to continue to play the game. Thus the higher I rate the game and the more I play the game the higher it is on my top ten list. Simple right?!

### š§® Bayesian Calculation #

Here is the formula used to calculate the Bayesian average for a game. It accounts for my game rating on BGG and the number of plays I’ve logged.

``````BA = R*(p+m)*M/(p+m)

R = rating for the game.
p = number of plays of the game.
m = minimum number of plays of the game.
M = mean rating for all games rated and owned.
``````

For the minimum number of play, I have initially set the value to five. As I log more game plays this value may be raised. As it stands now there is some wild variance in a few of my lower ranked games that causes them to appear higher on the top ten list than they should after a single play. With additional plays the list should even out over time. As for now I’m leaving the minimum number of plays set to five.

### š Python Script #

``````#!/usr/bin/env python
# encoding=utf8
"""Personal BGG Ratings."""

import argparse
import re
import sys
from functools import cmp_to_key
from operator import itemgetter

import requests
import requests_cache
import xmltodict

sys.getdefaultencoding()

def get_args():
"""Gather command line arguments or display help."""
parser = argparse.ArgumentParser(description='BGG Game Rankings',
default=argparse.SUPPRESS,
help='Show this help message and exit.')
'-v',
'--version',
action='version',
version='%(prog)s 5.0.0',
help="Show program's version number")
required=True, metavar='')
required=False, metavar='')
required=False, action='store_true')
return parser.parse_args()

def request_data(url):
"""Request data from boardgamegeek."""
requests_cache.install_cache('data_cache')
while True:
data = requests.get(url)
if not data.status_code == 200 or "try again later" in data.text:
continue
else:
break
return data.text

def bayesian_average(rating, mean, plays):
"""Bayesian average for game based on number of plays.

BA = R*(p+m)*M/(p+m)
R = rating for the game.
p = number of plays of the game.
m = minimum number of plays of the game.
M = mean rating for all games rated and owned.
"""
minimum_plays = 5
rating, mean, plays = float(rating), float(mean), float(plays)
bayes_avg = (rating * plays + minimum_plays * mean / plays + minimum_plays)
return bayes_avg

def multikeysort(items, columns):
"""Sort dictionary based on multiple keys."""
comparers = [((itemgetter(col[1:].strip()), 1) if col.startswith('-') else
(itemgetter(col.strip()), -1)) for col in columns]

def comparer(left, right):
for _fn, mult in comparers:
result = ((_fn(left) > _fn(right)) - (_fn(left) < _fn(right)))
if result:
return mult * result
return None

return sorted(items, key=cmp_to_key(comparer))

def calculate_mean(collection):
"""Calculate the mean ration for collection."""
ratings = []
for game in collection['items']['item']:
ratings.append(float(game['stats']['rating']['@value']))
mean = sum(ratings)/len(ratings)
return mean

"""Get user's collection from BGG."""
collection = []
baseurl = 'https://www.boardgamegeek.com/xmlapi2/'
'&rated=1&played=1&stats=1')
data = request_data(url)
doc = xmltodict.parse(data)
mean_rating = calculate_mean(doc)
for game in doc['items']['item']:
title = game['name']['#text'].strip()
player_rating = game['stats']['rating']['@value']
plays = game['numplays']
rating = bayesian_average(player_rating, mean_rating, plays)
collection.append({'name': title, 'player_rate': player_rating,
'rating': rating, 'plays': plays})

collection = multikeysort(collection, ['rating', 'name'])

return collection

def display_top_games(collection, count, detailed):
"""Display top games based on ratings then number of plays."""
if detailed:
print(f"{'Rank':<5}{'Rating':<7}{'Weighted':<10}{'Plays':<7}" \
f"{'Game':<100}")
else:
print(f"{'Rank':<5}{'Game':<100}")
rank = 1
rgx = re.compile('[%s]' % 'b\'\"')
for game in collection:
if detailed:
print(f"{rank:<5d}{float(game['player_rate']):<7.1f}" \
f"{float(game['rating']):<10.4f}" \
f"{game['plays']:<7}{rgx.sub('',game['name']):<100s}")
else:
print(f"{rank:<5d}{rgx.sub('',game['name']):<100s}")
if count:
if rank < int(count):
rank += 1
else:
sys.exit()
else:
rank += 1

if __name__ == "__main__":
ARGS = get_args()
display_top_games(get_collection(ARGS.user), ARGS.count, ARGS.detailed)
``````

### š requirements.txt #

``````requests
xmltodict
requests_cache
``````

### š³ Dockerfile #

``````FROM python:3-alpine

WORKDIR /usr/src/app

COPY requirements.txt ./
RUN pip install --no-cache-dir -r requirements.txt

COPY . .

ENTRYPOINT ["python", "./personal_game_ratings.py"]
CMD ["-h"]
``````