def filterAndStore(self, oldRule=None):
#print 'hola, hola'
self.examples = self.filter(self.data) # set examples
self.classifier = self.learner(self.examples) # set classifier
distribution = [0.0]* len(self.data.domain.classVar.values)
self.complexity = len(self.filter.conditions) # set rule complexity
self.ruleString = self.ruleToString()
self.orderedRuleString = self.orderedRuleToString()
if len(self.examples)>0:
for d in self.examples:
distribution[int(d.getclass())]+=1
distribution = map (lambda d: d/len(self.examples), distribution)
self.classDistribution = orange.Distribution(distribution) # set distribution
self.TP = filter(lambda e: e.getclass()==self.targetClass, self.examples) # True positives
self.FP = filter(lambda e: e.getclass()!=self.targetClass, self.examples) # flase positives
self.N_examples = len(self.data)
self.NTC = int(orange.getClassDistribution(self.data)[self.targetClass])
if oldRule == None or oldRule.complexity == 0:
self.P = len(self.data.filter({self.data.domain.classVar : self.targetClass}))
self.N = len(self.data.filter({self.data.domain.classVar : self.targetClass}, negate=1))
else:
self.P = len(oldRule.TP)
self.N = len(oldRule.FP)
self.TPlen = len(self.TP) * 1.0
self.FPlen = len(self.FP) * 1.0
self.calculateRefinementQuality()
self.calculateSelectionQuality()
self.calculateWeightedSelectionQuality()
self.quality = self.TPlen / (len(self.FP) + self.g) # set rule quality: generalization quocient
#print self.quality
self.support = 1.0* len(self.examples)/len(self.data) # set rule support
self.confidence = self.TPlen/len(self.examples)
else:
self.classDistribution = distribution
self.TP= []
self.FP= []
self.quality = 0 # set rule quality: generalization kvocient
self.support = 0 # set rule support
self.TPlen = 0
self.FPlen = 0
self.N_examples = 0
self.NTC = 0
self.refinement_quality = 0.0
self.selection_quality = 0.0
self.weighted_selection_quality = 0.0
def listOfClusters(root):
l = []
listOfClusters0(root, l)
return l
prune(root, 1.4)
for n, cluster in enumerate(listOfClusters(root)):
print "\n\n*** Cluster %i ***\n" % n
for ex in cluster:
print ex
for cluster in listOfClusters(root):
dist = orange.getClassDistribution(cluster)
for e, d in enumerate(dist):
print "%s: %3.0f " % (data.domain.classVar.values[e], d),
print
def listOfClustersT0(cluster, alist):
if not cluster.branches:
alist.append(orange.ExampleTable(cluster))
else:
for branch in cluster.branches:
listOfClustersT0(branch, alist)
def listOfClustersT(root):
l = []
indices2.randomGenerator = None
indices2.randseed = 42
for i in range(5):
print indices2(data)
print "\nIndices with p0 set as probability (not 'a number of')"
indices2.p0 = 0.25
print indices2(data)
print "\n... with stratification"
indices2.stratified = indices2.Stratified
ind = indices2(data)
print ind
data2 = data.select(ind)
od = orange.getClassDistribution(data)
sd = orange.getClassDistribution(data2)
od.normalize()
sd.normalize()
print od
print sd
print "\n... and without stratification"
indices2.stratified = indices2.NotStratified
print indices2(data)
ind = indices2(data)
print ind
data2 = data.select(ind)
od = orange.getClassDistribution(data)
sd = orange.getClassDistribution(data2)
od.normalize()
random.seed(0)
data = orange.ExampleTable("lenses")
id = -42
# Note that this is wrong. Id should be assigned by
# id = orange.newmetaid()
# We only do this so that the script gives the same output each time it's run
for example in data:
example[id] = orange.Value(random.random())
print data[0]
print orange.getClassDistribution(data)
print orange.getClassDistribution(data, id)
w = orange.FloatVariable("w")
data.domain.addmeta(id, w)
print data[0]
print data[0][id]
print data[0][w]
print data[0]["w"]
data[0][id] = orange.Value(w, 2.0)
data[0][id] = "2.0"
data[0][id] = 2.0
listOfClusters0(branch, alist)
def listOfClusters(root):
l = []
listOfClusters0(root, l)
return l
prune(root, 1.4)
for n, cluster in enumerate(listOfClusters(root)):
print "\n\n*** Cluster %i ***\n" % n
for ex in cluster:
print ex
for cluster in listOfClusters(root):
dist = orange.getClassDistribution(cluster)
for e, d in enumerate(dist):
print "%s: %3.0f " % (data.domain.classVar.values[e], d),
print
def listOfClustersT0(cluster, alist):
if not cluster.branches:
alist.append(orange.ExampleTable(cluster))
else:
for branch in cluster.branches:
listOfClustersT0(branch, alist)
def listOfClustersT(root):
l = []
listOfClustersT0(root, l)
return l
print "\nIndices with randseed"
indices2.randomGenerator = None
indices2.randseed = 42
for i in range(5):
print indices2(data)
print "\nIndices with p0 set as probability (not 'a number of')"
indices2.p0 = 0.25
print indices2(data)
print "\n... with stratification"
indices2.stratified = indices2.Stratified
ind = indices2(data)
print ind
data2 = data.select(ind)
od = orange.getClassDistribution(data)
sd = orange.getClassDistribution(data2)
od.normalize()
sd.normalize()
print od
print sd
print "\n... and without stratification"
indices2.stratified = indices2.NotStratified
print indices2(data)
ind = indices2(data)
print ind
data2 = data.select(ind)
od = orange.getClassDistribution(data)
sd = orange.getClassDistribution(data2)
od.normalize()