'1001', '1010', '1011', '1100', '1101', '1110', '1111'
]
nPatterns = len(patterns)
for formatType in formats:
patternFreqs = dict(zip(patterns, [0 for i in range(nPatterns)]))
filename = 'Brackets/{0}/allBrackets{0}.json'.format(formatType)
with open(filename, 'r') as inputFile:
jsonData = inputFile.read().replace('\n', '')
jsonToPython = json.loads(jsonData)
bracketList = jsonToPython['brackets']
numBrackets = len(bracketList)
for i in range(numBrackets):
bracketDict = bracketList[i]['bracket']
bracket = buildBracketFromJson(bracketDict)
offset = 15
finalGames = ''
for region in range(4):
finalGames += bracket.fullVector[14 + region * offset]
patternFreqs[finalGames] = patternFreqs[finalGames] + 1
# rowSums = [patternFreqs['000'] + patternFreqs['001'], patternFreqs['010'] + patternFreqs['011'], patternFreqs['100'] + patternFreqs['101'], patternFreqs['110'] + patternFreqs['111']]
# colSums = [patternFreqs['000'] + patternFreqs['010'] + patternFreqs['100'] + patternFreqs['110'], patternFreqs['001'] + patternFreqs['011'] + patternFreqs['101'] + patternFreqs['111']]
# nObservations = numBrackets * numRegions
print '{0}: Regional Finals GOF Test'.format(formatType)
print patternFreqs
expFreq = numBrackets / nPatterns
chiSquare = 0
region = pos / 15
offset = 15 * region
posMod15 = pos - offset
# Only looking at R64 and R32 positions
if region >= 4 or posMod15 >= 12:
continue
print 'Position:,{0}'.format(pos)
if posMod15 > 7:
freqs_1 = dict(zip(patterns, zeroList)) # R32 vs. S16E8 triplet
triplet_1 = getS16E8Positions(pos)
for i in range(numBrackets):
bracket = buildBracketFromJson(bracketList[i]['bracket'])
bracketString = bracket.fullVector
string_1 = '{0}{1}{2}'.format(bracketString[triplet_1[0]],
bracketString[triplet_1[1]],
bracketString[triplet_1[2]])
string_1 = '{0}_{1}'.format(str(bracketString[pos]), string_1)
freqs_1[string_1] += 1
print 'R32 vs. S16/E8 Triplet:'
printTable(pos, triplet_1, freqs_1)
continue
freqs_2 = dict(zip(patterns, zeroList)) # R64 vs. R32->S16->E8 path
def testPairwiseIndependence(pos1, pos2, formatType, isPooled=False):
patterns = ['00', '01', '10', '11']
patternFreqs = [0 for i in range(4)]
filename = 'Brackets/{0}/allBrackets{0}.json'.format(formatType)
with open(filename, 'r') as inputFile:
jsonData = inputFile.read().replace('\n', '')
jsonToPython = json.loads(jsonData)
bracketList = jsonToPython['brackets']
numBrackets = len(bracketList)
for i in range(numBrackets):
bracketDict = bracketList[i]['bracket']
bracket = buildBracketFromJson(bracketDict)
nRegions = 1
if isPooled:
nRegions = 4
for region in range(nRegions):
offset = region * 15
pos1Result = int(bracket.fullVector[pos1 + offset])
pos2Result = int(bracket.fullVector[pos2 + offset])
index = pos1Result * 2 + pos2Result
patternFreqs[index] = patternFreqs[index] + 1
nObservations = numBrackets * nRegions
n11 = patternFreqs[3]
n10 = patternFreqs[2]
n01 = patternFreqs[1]
n00 = patternFreqs[0]
n1x = n11 + n10
n0x = n01 + n00
nx1 = n11 + n01
nx0 = n10 + n00
phiNum = n11 * n00 - n10 * n01
phiDen = sqrt(n1x * n0x * nx1 * nx0)
if phiDen > 0:
phi = phiNum / phiDen
else:
phi = 0
chiSquare = phi * phi * nObservations
summaryString = '{0:02d},{1:02d},{2},{3},{4},{5},{6:.2f},{7:.2f}'.format(
pos1, pos2, n11, n10, n01, n00, phi, chiSquare)
# The chi-square critical value for 1 degree of freedom and alpha = 0.05
# is 3.841. (Source: http://www.itl.nist.gov/div898/handbook/eda/section3/eda3674.htm)
# We print the result only if it is deemed significant.
isSignificant = 0
if chiSquare >= 3.841:
isSignificant = 1
print summaryString
return isSignificant
def testTwoToOneIndependence(pos1, pos2, pos3, isPooled=False):
formats = ['TTT', 'TTF', 'TFT', 'TFF', 'FTT', 'FTF', 'FFT', 'FFF']
patterns = ['000', '001', '010', '011', '100', '101', '110', '111']
numRegions = 4
if not isPooled:
numRegions = 1
print 'formatType, pos1, pos2, pos3, chiSquared, n111, n110, n101, n100, n011, n010, n001, n000'
for formatType in formats:
patternFreqs = dict(zip(patterns, [0 for i in range(8)]))
filename = 'Brackets/{0}/allBrackets{0}.json'.format(formatType)
with open(filename, 'r') as inputFile:
jsonData = inputFile.read().replace('\n', '')
jsonToPython = json.loads(jsonData)
bracketList = jsonToPython['brackets']
numBrackets = len(bracketList)
for i in range(numBrackets):
bracketDict = bracketList[i]['bracket']
bracket = buildBracketFromJson(bracketDict)
for region in range(numRegions):
offset = region * 15
threeGames = ''
if pos1 + offset >= len(bracket.fullVector):
print 'Houston, we have a problem. pos1 + offset = {0}'.format(
pos1 + offset)
threeGames += bracket.fullVector[pos1 + offset]
threeGames += bracket.fullVector[pos2 + offset]
threeGames += bracket.fullVector[pos3 + offset]
patternFreqs[threeGames] = patternFreqs[threeGames] + 1
n000 = patternFreqs['000']
n001 = patternFreqs['001']
n010 = patternFreqs['010']
n011 = patternFreqs['011']
n100 = patternFreqs['100']
n101 = patternFreqs['101']
n110 = patternFreqs['110']
n111 = patternFreqs['111']
nxx1 = n001 + n011 + n101 + n111
nxx0 = n000 + n010 + n100 + n110
n11x = n111 + n110
n10x = n101 + n100
n01x = n011 + n010
n00x = n001 + n000
nObservations = numBrackets * numRegions
rowSums = [
patternFreqs['000'] + patternFreqs['001'],
patternFreqs['010'] + patternFreqs['011'],
patternFreqs['100'] + patternFreqs['101'],
patternFreqs['110'] + patternFreqs['111']
]
colSums = [
patternFreqs['000'] + patternFreqs['010'] + patternFreqs['100'] +
patternFreqs['110'], patternFreqs['001'] + patternFreqs['011'] +
patternFreqs['101'] + patternFreqs['111']
]
chiSquare = 0
for r in range(len(rowSums)):
for c in range(len(colSums)):
expFreq = rowSums[r] * colSums[c] * 1.0 / nObservations
if expFreq == 0: # This hack dodges division by zero
expFreq = 0.00000001
obsFreq = patternFreqs[patterns[2 * r + c]] * 1.0
chiSquare += (obsFreq - expFreq)**2 / expFreq
summaryString = '{12},{0:02d},{1:02d},{2:02d},{3:.3f},{4},{5},{6},{7},{8},{9},{10},{11}'.format(
pos1, pos2, pos3, chiSquare, n111, n110, n101, n100, n011, n010,
n001, n000, formatType)
# if chiSquare >= 7.815: # critical value for 3 degrees of freedom, alpha = 0.05
print summaryString