def createBayesianNetwork():
learner=gum.BNLearner("logs/Log/WholeLog.csv")
learner.useK2([0,1,2,3,4,5,6,7,8,9,10,11,12])
bn2=learner.learnBN()
print("Learned in {0}s".format(learner.currentTime()))
gnb.showBN(bn2)
return bn2
def createBayesianNetwork():
learner = gum.BNLearner("logs/Log/WholeLog.csv")
learner.useK2([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12])
bn2 = learner.learnBN()
print("Learned in {0}s".format(learner.currentTime()))
gnb.showBN(bn2)
return bn2
df = pandas.read_csv(os.path.join('res', 'titanic', '/content/gdrive/My Drive/train_data2.csv'))
for k in train_df.keys():
print('{0}: {1}'.format(k, len(train_df[k].unique())))
template=gum.BayesNet()
template.add(gum.LabelizedVariable("target", "target", ['<=50K', '>50K']))
template.add(gum.LabelizedVariable("sex", "sex",['Male','Female']))
template.add(gum.LabelizedVariable("age_range", "age_range",['0-20','21-30','31-65','66-90']))
template.add(gum.LabelizedVariable("race", "race",['White', 'Asian-Pac-Islander', 'Amer-Indian-Eskimo', 'Other', 'Black']))
template.add(gum.LabelizedVariable("workclass", "workclass",['Private', 'Self-emp-not-inc', 'Self-emp-inc', 'Federal-gov', 'Local-gov', 'State-gov', 'Without-pay', 'Never-worked']))
template.add(gum.LabelizedVariable("relationship", "relationship", ['Wife', 'Own-child', 'Husband', 'Not-in-family', 'Other-relative', 'Unmarried']))
template.add(gum.LabelizedVariable("marital_status", "marital_status", ['Married-civ-spouse', 'Divorced', 'Never-married', 'Separated', 'Widowed', 'Married-spouse-absent', 'Married-AF-spouse']))
template.add(gum.LabelizedVariable("occupation", "occupation",['Tech-support', 'Craft-repair', 'Other-service', 'Sales', 'Exec-managerial', 'Prof-specialty', 'Handlers-cleaners', 'Machine-op-inspct', 'Adm-clerical', 'Farming-fishing', 'Transport-moving', 'Priv-house-serv', 'Protective-serv', 'Armed-Forces']))
gnb.showBN(template)
train_df.to_csv(os.path.join('/content/gdrive/My Drive/train_data2.csv'), index=False)
file = os.path.join('res', 'titanic', '/content/gdrive/My Drive/train_data2.csv')
learner = gum.BNLearner(file, template)
bn = learner.learnBN()
bn
gnb.showInformation(bn,{},size="20")
gnb.showInference(bn)
gnb.showPosterior(bn,evs={"sex": "Male", "age_range": '21-30'},target='target')
gnb.sideBySide(bn, gum.MarkovBlanket(bn, 'target'), captions=["Learned Bayesian Network", "Markov blanket of 'target'"])
# %% markdown # ## Parameter Learning from the database # We give the `asiaBN` bayesian network as a parameter for the learner in order to have the variables and order of labels for each variable. # %% codecell # using the BN as template for variables and labels learner = gum.BNLearner(outPath, asiaBN) learner.setInitialDAG(g = asiaBN.dag()) # Learn the parameters when structure is known: asiaBN_learnedParams: BayesNet = learner.learnParameters() gnb.showBN(asiaBN_learnedParams) # gnb.showBN(asiaBN) # same thing # %% codecell # This is the bad example: learning without the initial template gets the nodes and structure wrong learnerNoTemplate = gum.BNLearner(outPath) learnerNoTemplate.setInitialDAG(g = asiaBN.dag()) asiaBNNoTemplate: BayesNet = learnerNoTemplate.learnParameters() gnb.showBN(asiaBNNoTemplate) # %% codecell # This is what the DAG looks like asiaBN.dag() # %% codecell asiaBNNoTemplate.dag() # same # %% codecell
def trainmodel(filename):
ob, hb, lb, cb, vb = gentt(filename)
#print(ob,hb,lb,cb)
#build the model
bn = gum.BayesNet(filename)
Open0 = bn.add('Open0', ob)
High0 = bn.add('High0', hb)
Low0 = bn.add('Low0', lb)
Close0 = bn.add('Close0', cb)
Volume0 = bn.add('Volume0', vb)
Open1 = bn.add('Open1', ob)
High1 = bn.add('High1', hb)
Low1 = bn.add('Low1', lb)
Close1 = bn.add('Close1', cb)
Volume1 = bn.add('Volume1', vb)
learner = gum.BNLearner(filename + "_train.csv", bn)
learner.addForbiddenArc('Open1', 'Open0')
learner.addForbiddenArc('Open1', 'Close0')
learner.addForbiddenArc('Open1', 'High0')
learner.addForbiddenArc('Open1', 'Low0')
learner.addForbiddenArc('Open1', 'Volume0')
learner.addForbiddenArc('High1', 'Open0')
learner.addForbiddenArc('High1', 'Close0')
learner.addForbiddenArc('High1', 'High0')
learner.addForbiddenArc('High1', 'Low0')
learner.addForbiddenArc('High1', 'Volume0')
learner.addForbiddenArc('Low1', 'Open0')
learner.addForbiddenArc('Low1', 'Close0')
learner.addForbiddenArc('Low1', 'High0')
learner.addForbiddenArc('Low1', 'Low0')
learner.addForbiddenArc('Low1', 'Volume0')
learner.addForbiddenArc('Close1', 'Open0')
learner.addForbiddenArc('Close1', 'Close0')
learner.addForbiddenArc('Close1', 'High0')
learner.addForbiddenArc('Close1', 'Low0')
learner.addForbiddenArc('Close1', 'Volume0')
learner.addForbiddenArc('Volume1', 'Open0')
learner.addForbiddenArc('Volume1', 'Close0')
learner.addForbiddenArc('Volume1', 'High0')
learner.addForbiddenArc('Volume1', 'Low0')
learner.addForbiddenArc('Volume1', 'Volume0')
#learner.addMandatoryArc('Close0','Close1')
learner.useLocalSearchWithTabuList()
bn = learner.learnBN()
gnb.showBN(bn)
learner = gum.BNLearner(filename + "_train.csv", bn)
learner.setInitialDAG(bn.dag())
learner.useAprioriSmoothing(1)
bn = learner.learnParameters()
#gnb.showInference(bn,evs={})
#do inference and calculate the accuracy
ie = gum.LazyPropagation(bn)
ie.makeInference()
N = 0.0
acc = 0
with open(filename + '_test.csv', 'r', encoding="utf-8") as csvfile:
reader = csv.reader(csvfile)
for line in list(reader)[1:]:
c, o, h, l, v, t = [
line[0], line[1], line[2], line[3], line[4], line[5]
]
ie.eraseAllEvidence()
ie.setEvidence({
'Close0': c,
'Open0': o,
'High0': h,
'Low0': l,
'Volume0': v
})
ie.makeInference()
prob = ie.posterior(Close1).tolist()
if prob[0] < 0.6:
N = N + 1
if t == '1':
acc = acc + 1
return acc, N
def main():
bn = gum.BayesNet('nuc_inf')
#add variables to the network
va = gum.LabelizedVariable('nuc', 'a labelized variable', 2)
va.addLabel('-1')
nuc = bn.add(va)
A = bn.add('A', 6)
R, N = [bn.add(name, 7) for name in ['R', 'N']]
D, Q = [bn.add(name, 2) for name in ['D', 'Q']]
partition("protein")
learner = gum.BNLearner("protein_train.csv", bn)
#These arcs can be added or deleted
#learner.addMandatoryArc('A','nuc')
#learner.addMandatoryArc('R','nuc')
#learner.addMandatoryArc('Q','nuc')
#learner.addMandatoryArc('N','nuc')
#learner.addMandatoryArc('D','nuc')
learner.useLocalSearchWithTabuList()
bn0 = learner.learnBN()
gnb.showBN(bn0)
learner.useGreedyHillClimbing()
bn1 = learner.learnBN()
gnb.showBN(bn1)
learner.useK2([5, 4, 3, 2, 1, 0])
bn2 = learner.learnBN()
gnb.showBN(bn2)
#We have 2 different BN structures according to the previous parts. Now, we do parameter learning
learner = gum.BNLearner("protein_train.csv", bn)
learner.setInitialDAG(bn0.dag())
learner.useAprioriSmoothing(1)
bn01 = learner.learnParameters() #first
gnb.showBN(bn01)
learner = gum.BNLearner("protein_train.csv", bn)
learner.setInitialDAG(bn2.dag())
learner.useAprioriSmoothing(1)
bn11 = learner.learnParameters() #second
gnb.showBN(bn11)
#first
ie1 = gum.LazyPropagation(bn01)
ie1.makeInference()
gnb.showInference(bn01, evs={})
#second
ie2 = gum.LazyPropagation(bn11)
ie2.makeInference()
gnb.showInference(bn11, evs={})
with open('protein_test.csv', 'r', encoding="utf-8") as csvfile:
reader = csv.reader(csvfile)
count1 = 1
count2 = 1
acc1 = 0
acc2 = 0
for line in list(reader)[1:]:
vnuc, vA, vR, vN, vD, vQ = [
int(line[0]),
int(line[1]),
int(line[2]),
int(line[3]),
int(line[4]),
int(line[5])
]
#print(vnuc,vA,vR,vN,vD,vQ)
ie2.eraseAllEvidence()
ie1.eraseAllEvidence()
ie1.setEvidence({'A': vA, 'R': vR, 'N': vN, 'D': vD, 'Q': vQ})
ie2.setEvidence({'A': vA, 'R': vR, 'N': vN, 'D': vD, 'Q': vQ})
ie1.makeInference()
ie2.makeInference()
ie2.addTarget(nuc)
ie1.addTarget(nuc)
if len(ie2.posterior(nuc).argmax()
) == 1: #if we have one determined value of prob
#print(ie2.posterior(nuc))
#print(ie2.posterior(nuc).argmax()[0]['nuc'])
if ie2.posterior(nuc).argmax()[0]['nuc'] == 2: #nuc=-1
if vnuc == -1:
acc2 = acc2 + 1
if ie2.posterior(nuc).argmax()[0]['nuc'] == vnuc:
acc2 = acc2 + 1
count2 = count2 + 1
if len(ie1.posterior(nuc).argmax()) == 1:
#print(ie1.posterior(nuc))
#print(ie1.posterior(nuc).argmax()[0]['nuc'])
if ie1.posterior(nuc).argmax()[0]['nuc'] == 2:
if vnuc == -1:
acc1 = acc1 + 1
if ie1.posterior(nuc).argmax()[0]['nuc'] == vnuc:
acc1 = acc1 + 1
count1 = count1 + 1
acc2 = acc2 / count2
acc1 = acc1 / count1
print(acc2, acc1)