def GetRewardWeights(M, Rep=-1, alpha=Decimal('0.1')):
if Rep==-1:
Rep=DemocracyRep(M)
if type(Rep[0])==list:
Rep=map(lambda x: x[0], Rep)
Results=custommath.WeightedPrinComp(M, Rep)
FirstLoading=Results['Loadings']
FirstScore=Results['Scores']
a=min(map(abs,FirstScore))
b=max(FirstScore)
Set1=map(lambda x: x+a, FirstScore)
Set2=map(lambda x: x-b, FirstScore)
Old=custommath.dot([Rep], M)
New1=custommath.dot([custommath.GetWeight(Set1)], M)
New2=custommath.dot([custommath.GetWeight(Set2)], M)
def sub_f(a, b): return a-b
def f(n): return sum(map(lambda x: x**2, map(sub_f, n[0], Old[0])))
RefInd=f(New1)-f(New2)
if(RefInd<=0): AdjPrinComp = Set1
else: AdjPrinComp = Set2
RowRewardWeighted=Rep
if max(map(abs,AdjPrinComp))!=0:
m=custommath.mean(Rep)
RRW=[]
for i in range(len(Rep)):
RRW.append(AdjPrinComp[i]*Rep[i]/m)
RowRewardWeighted=custommath.GetWeight(RRW)
SmoothedR=[]
for i in range(len(Rep)):
SmoothedR.append(alpha*RowRewardWeighted[i]+(1-alpha)*Rep[i])
Out = {"FirstL":FirstLoading, "OldRep":Rep, "ThisRep":RowRewardWeighted, "SmoothRep":SmoothedR}
return(Out)
def Factory(M0, Rep, CatchP=Decimal('0.1'), MaxRow=5000):
Rep=custommath.ReWeight(Rep)
Filled=FillNa(M0, Rep, CatchP)
PlayerInfo=GetRewardWeights(Filled, Rep, Decimal('0.1'))
AdjLoadings=PlayerInfo['FirstL']
#print('smoothrep: ' +str(PlayerInfo['SmoothRep']))#way wrong
#print('Filled: ' +str(Filled))
DecisionOutcomesRaw=custommath.dot([PlayerInfo['SmoothRep']], Filled)[0]
#print('raw outcomes: ' +str(DecisionOutcomesRaw))
DecisionOutcomeAdj=map(lambda x: custommath.Catch(x, CatchP), DecisionOutcomesRaw)
Certainty=map(lambda x: 2*(x-Decimal('0.5')), DecisionOutcomesRaw)
Certainty=map(abs, Certainty)
ConReward=custommath.GetWeight(Certainty)
Avg_Certainty=custommath.mean(Certainty)
DecisionOutcomeAdj=[]
for i, raw in enumerate(DecisionOutcomesRaw):
DecisionOutcomeAdj.append(custommath.Catch(raw, CatchP))
DecisionOutcomeFinal=DecisionOutcomeAdj
NAmat=map(lambda row: map(lambda x: 1 if type(x) in [str, unicode] else 0, row), M0)
a=custommath.dot([PlayerInfo['SmoothRep']], NAmat)[0]
ParticipationC=map(lambda x: 1-x, a)
v=map(sum, NAmat)
ParticipationR=map(lambda x: 1-x/Decimal(len(NAmat[0])), v)
PercentNA=1-custommath.mean(ParticipationC)
NAbonusR = custommath.GetWeight(ParticipationR)
def plus(a, b): return a+b
RowBonus = map(plus, map(lambda x: x*PercentNA, NAbonusR), map(lambda x: x*(1-PercentNA), PlayerInfo['SmoothRep']))
NAbonusC=custommath.GetWeight(ParticipationC)
ColBonus=map(plus, map(lambda x: x*PercentNA, NAbonusC), map(lambda x: x*(1-PercentNA), ConReward))
namatsum=[]
for i in range(len(NAmat[0])):
namatsum.append(sum(map(lambda x: x[i], NAmat)))
Output = {
'Original': M0,
'Filled': Filled,
'Agents': {
'OldRep': PlayerInfo['OldRep'],#[0],
'ThisRep': PlayerInfo['ThisRep'],
'SmoothRep': PlayerInfo['SmoothRep'],
'NArow': map(sum, NAmat),#.sum(axis=1).base,
'ParticipationR': ParticipationR,#.base,
'RelativePart': NAbonusR,#.base,
'RowBonus': RowBonus,#.base,
},
'Decisions': {
'First Loading': AdjLoadings,
'DecisionOutcomes_Raw': DecisionOutcomesRaw,
'Consensus Reward': ConReward,
'Certainty': Certainty,
'NAs Filled': namatsum,#NAmat.sum(axis=0),
'ParticipationC': ParticipationC,
'Author Bonus': ColBonus,
'DecisionOutcome_Final': DecisionOutcomeFinal,
},
'Participation': 1 - PercentNA,
'Certainty': Avg_Certainty,
}
return Output
def FillNa(Mna, Rep, Catchp=Decimal('0.1')):
Mnew = Mna
MnewC = Mna
if any_NA(Mna):
DecisionOutcomesRaw = GetDecisionOutcomes(Mna, Rep)
NAmat = map(lambda row: map(lambda x: 1 if x == 'NA' else 0, row), Mna)
Mnew = map(lambda row: map(lambda x: 0 if x == 'NA' else x, row), Mna)
NAsToFill = custommath.dot(NAmat, diag(DecisionOutcomesRaw))
for row in range(len(Mnew)):
for i in range(len(Mnew[row])):
Mnew[row][i] = Mnew[row][i] + NAsToFill[row][i]
return (map(
lambda row: map(lambda x: custommath.Catch(x, Catchp), row), Mnew))
return (Mna)
def GetDecisionOutcomes(Mtemp, Rep):
# Determines the Outcomes of Decisions based on the provided reputation (weighted vote)
#For each column
out = []
for i in range(len(Mtemp[0])):
Row = []
Col = []
c = map(lambda x: x[i], Mtemp)
for j in range(len(c)):
if type(c[j]) not in [str, unicode]:
Row.append(Rep[j])
Col.append(c[j])
Row = custommath.ReWeight(Row)
out.append(v_dot(Col, Row))
return custommath.switch_row_cols(out)
def FillNa(Mna, Rep, Catchp=Decimal('0.1')):
Mnew = Mna
MnewC=Mna
if any_NA(Mna):
DecisionOutcomesRaw=GetDecisionOutcomes(Mna, Rep)
NAmat=map(lambda row: map(lambda x: 1 if x=='NA' else 0, row), Mna)
Mnew=map(lambda row: map(lambda x: 0 if x=='NA' else x, row), Mna)
NAsToFill=custommath.dot(NAmat, diag(DecisionOutcomesRaw))
for row in range(len(Mnew)):
for i in range(len(Mnew[row])):
Mnew[row][i]=Mnew[row][i]+NAsToFill[row][i]
return(map(lambda row: map(lambda x: custommath.Catch(x, Catchp), row), Mnew))
return(Mna)
def GetDecisionOutcomes(Mtemp, Rep):
# Determines the Outcomes of Decisions based on the provided reputation (weighted vote)
#For each column
out=[]
for i in range(len(Mtemp[0])):
Row=[]
Col=[]
c=map(lambda x: x[i], Mtemp)
for j in range(len(c)):
if type(c[j]) not in [str, unicode]:
Row.append(Rep[j])
Col.append(c[j])
Row=custommath.ReWeight(Row)
out.append(v_dot(Col, Row))
return custommath.switch_row_cols(out)
def GetRewardWeights(M, Rep=-1, alpha=Decimal('0.1')):
if Rep == -1:
Rep = DemocracyRep(M)
if type(Rep[0]) == list:
Rep = map(lambda x: x[0], Rep)
Results = custommath.WeightedPrinComp(M, Rep)
FirstLoading = Results['Loadings']
FirstScore = Results['Scores']
a = min(map(abs, FirstScore))
b = max(FirstScore)
Set1 = map(lambda x: x + a, FirstScore)
Set2 = map(lambda x: x - b, FirstScore)
Old = custommath.dot([Rep], M)
New1 = custommath.dot([custommath.GetWeight(Set1)], M)
New2 = custommath.dot([custommath.GetWeight(Set2)], M)
def sub_f(a, b):
return a - b
def f(n):
return sum(map(lambda x: x**2, map(sub_f, n[0], Old[0])))
RefInd = f(New1) - f(New2)
if (RefInd <= 0): AdjPrinComp = Set1
else: AdjPrinComp = Set2
RowRewardWeighted = Rep
if max(map(abs, AdjPrinComp)) != 0:
m = custommath.mean(Rep)
RRW = []
for i in range(len(Rep)):
RRW.append(AdjPrinComp[i] * Rep[i] / m)
RowRewardWeighted = custommath.GetWeight(RRW)
SmoothedR = []
for i in range(len(Rep)):
SmoothedR.append(alpha * RowRewardWeighted[i] + (1 - alpha) * Rep[i])
Out = {
"FirstL": FirstLoading,
"OldRep": Rep,
"ThisRep": RowRewardWeighted,
"SmoothRep": SmoothedR
}
return (Out)
def DemocracyRep(x):
v = []
for i in range(len(x)):
v.append(1)
return (custommath.ReWeight(v))
def Factory(M0, Rep, CatchP=Decimal('0.1'), MaxRow=5000):
Rep = custommath.ReWeight(Rep)
Filled = FillNa(M0, Rep, CatchP)
PlayerInfo = GetRewardWeights(Filled, Rep, Decimal('0.1'))
AdjLoadings = PlayerInfo['FirstL']
#print('smoothrep: ' +str(PlayerInfo['SmoothRep']))#way wrong
#print('Filled: ' +str(Filled))
DecisionOutcomesRaw = custommath.dot([PlayerInfo['SmoothRep']], Filled)[0]
#print('raw outcomes: ' +str(DecisionOutcomesRaw))
DecisionOutcomeAdj = map(lambda x: custommath.Catch(x, CatchP),
DecisionOutcomesRaw)
Certainty = map(lambda x: 2 * (x - Decimal('0.5')), DecisionOutcomesRaw)
Certainty = map(abs, Certainty)
ConReward = custommath.GetWeight(Certainty)
Avg_Certainty = custommath.mean(Certainty)
DecisionOutcomeAdj = []
for i, raw in enumerate(DecisionOutcomesRaw):
DecisionOutcomeAdj.append(custommath.Catch(raw, CatchP))
DecisionOutcomeFinal = DecisionOutcomeAdj
NAmat = map(
lambda row: map(lambda x: 1
if type(x) in [str, unicode] else 0, row), M0)
a = custommath.dot([PlayerInfo['SmoothRep']], NAmat)[0]
ParticipationC = map(lambda x: 1 - x, a)
v = map(sum, NAmat)
ParticipationR = map(lambda x: 1 - x / Decimal(len(NAmat[0])), v)
PercentNA = 1 - custommath.mean(ParticipationC)
NAbonusR = custommath.GetWeight(ParticipationR)
def plus(a, b):
return a + b
RowBonus = map(plus, map(lambda x: x * PercentNA, NAbonusR),
map(lambda x: x * (1 - PercentNA), PlayerInfo['SmoothRep']))
NAbonusC = custommath.GetWeight(ParticipationC)
ColBonus = map(plus, map(lambda x: x * PercentNA, NAbonusC),
map(lambda x: x * (1 - PercentNA), ConReward))
namatsum = []
for i in range(len(NAmat[0])):
namatsum.append(sum(map(lambda x: x[i], NAmat)))
Output = {
'Original': M0,
'Filled': Filled,
'Agents': {
'OldRep': PlayerInfo['OldRep'], #[0],
'ThisRep': PlayerInfo['ThisRep'],
'SmoothRep': PlayerInfo['SmoothRep'],
'NArow': map(sum, NAmat), #.sum(axis=1).base,
'ParticipationR': ParticipationR, #.base,
'RelativePart': NAbonusR, #.base,
'RowBonus': RowBonus, #.base,
},
'Decisions': {
'First Loading': AdjLoadings,
'DecisionOutcomes_Raw': DecisionOutcomesRaw,
'Consensus Reward': ConReward,
'Certainty': Certainty,
'NAs Filled': namatsum, #NAmat.sum(axis=0),
'ParticipationC': ParticipationC,
'Author Bonus': ColBonus,
'DecisionOutcome_Final': DecisionOutcomeFinal,
},
'Participation': 1 - PercentNA,
'Certainty': Avg_Certainty,
}
return Output
