# Description: Uses cross-validation to compare regression tree and k-nearest neighbors
# Category: modelling, evaluation
# Uses: housing
# Classes: orngStat.MSE, orngTest.crossValidation, MajorityLearner, orngTree.TreeLearner, orange.kNNLearner
# Referenced: regression.htm
import orange, orngTree, orngTest, orngStat
data = orange.ExampleTable("housing.tab")
maj = orange.MajorityLearner()
maj.name = "default"
rt = orngTree.TreeLearner(measure="retis", mForPruning=2, minExamples=20)
rt.name = "regression tree"
k = 5
knn = orange.kNNLearner(k=k)
knn.name = "k-NN (k=%i)" % k
learners = [maj, rt, knn]
data = orange.ExampleTable("housing.tab")
results = orngTest.crossValidation(learners, data, folds=10)
mse = orngStat.MSE(results)
print "Learner MSE"
for i in range(len(learners)):
print "%-15s %5.3f" % (learners[i].name, mse[i])
def __call__(self, data, y=None, x=None, nc=None, weight=None, **kwds):
if y == None:
y = [data.domain.classVar]
if x == None:
x = [v for v in data.domain.variables if v not in y]
Ncomp = nc if nc is not None else len(x)
dataX = orange.ExampleTable(orange.Domain(x, False), data)
dataY = orange.ExampleTable(orange.Domain(y, False), data)
# transformation to numpy arrays
X = dataX.toNumpy()[0]
Y = dataY.toNumpy()[0]
# data dimensions
n, mx = numpy.shape(X)
my = numpy.shape(Y)[1]
# Z-scores of original matrices
YMean = numpy.mean(Y, axis=0)
YStd = numpy.std(Y, axis=0)
XMean = numpy.mean(X, axis=0)
XStd = numpy.std(X, axis=0)
#FIXME: standard deviation should never be 0. Ask Lan, if the following
#fix is ok.
XStd = numpy.maximum(XStd, 10e-16)
YStd = numpy.maximum(YStd, 10e-16)
X = (X - XMean) / XStd
Y = (Y - YMean) / YStd
P = numpy.empty((mx, Ncomp))
C = numpy.empty((my, Ncomp))
T = numpy.empty((n, Ncomp))
U = numpy.empty((n, Ncomp))
B = numpy.zeros((Ncomp, Ncomp))
W = numpy.empty((mx, Ncomp))
E, F = X, Y
# main algorithm
for i in range(Ncomp):
u = numpy.random.random_sample((n, 1)) #FIXME random seed?
w = normalize(dot(E.T, u))
t = normalize(dot(E, w))
dif = t
# iterations for loading vector t
while numpy.linalg.norm(dif) > 10e-16:
c = normalize(dot(F.T, t))
u = dot(F, c)
w = normalize(dot(E.T, u))
t0 = normalize(dot(E, w))
dif = t - t0
t = t0
#print "T", T
#print "X*W", numpy.dot(X,W)
T[:, i] = t.T
U[:, i] = u.T
C[:, i] = c.T
W[:, i] = w.T
b = dot(t.T, u)[0]
B[i][i] = b
p = dot(E.T, t)
P[:, i] = p.T
E = E - dot(t, p.T)
F = F - b * dot(t, c.T)
# esimated Y
YE = dot(dot(T, B), C.T) * YStd + YMean
Y = Y * numpy.std(Y, axis=0) + YMean
BPls = dot(dot(numpy.linalg.pinv(P.T), B), C.T)
partial = {}
if self.save_partial:
partial["T"] = T
partial["U"] = U
partial["C"] = C
partial["W"] = W
partial["P"] = P
return PLSRegression(domain=data.domain,
BPls=BPls,
YMean=YMean,
YStd=YStd,
XMean=XMean,
XStd=XStd,
name=self.name,
**partial)
self.progressBarFinished()
self.classifier.name = self.LearnerName
self.classifier.setattr("data", self.data)
if self.data.domain.classVar:
self.send("Classifier", self.classifier)
self.send("SOM", self.classifier)
def sendReport(self):
self.reportSettings(
"Topology",
[("Shape", ["hexagonal", "rectangular"][self.topology]),
("Size", "%i columns, %i rows" % (self.xdim, self.ydim))])
self.reportSettings(
"Optimization",
[("Initialization", ["linear", "random"][self.initialization]),
("Neighborhood", ["Gaussian", "bubble"][self.neighborhood]),
("Radius", "initial: %i, final: %i" %
(self.radius1, self.radius2)),
("Number of iterations", self.iterations1)])
if __name__ == "__main__":
app = QApplication(sys.argv)
w = OWSOM()
## app.setMainWidget(w)
w.show()
data = orange.ExampleTable("../../doc/datasets/iris.tab")
w.setData(data)
app.exec_()
# Description: Creates a list of association rules, selects five rules and prints them out
# Category: description
# Uses: imports-85
# Classes: orngAssoc.build, Preprocessor_discretize, EquiNDiscretization
# Referenced: assoc.htm
import orange, orngAssoc
data = orange.ExampleTable("imports-85")
data = orange.Preprocessor_discretize(data, \
method=orange.EquiNDiscretization(numberOfIntervals=3))
data = data.select(range(10))
rules = orange.AssociationRulesInducer(data, support=0.4)
print "%i rules with support higher than or equal to %5.3f found.\n" % (len(rules), 0.4)
orngAssoc.sort(rules, ["support", "confidence"])
orngAssoc.printRules(rules[:5], ["support", "confidence"])
print
del rules[:3]
orngAssoc.printRules(rules[:5], ["support", "confidence"])
print
def __call__(self, newRule, examples, weightID, targetClass, prior):
N = len(examples)
ny = len(filter(lambda e: e.getclass() == targetClass, examples))
N1 = n1x = n1xy = 0
for e in examples:
tmp = e.getweight(weightID)
N1 += tmp
if newRule.filter(e):
n1x += tmp
if e.getclass() == targetClass:
n1xy += tmp
wracc = n1xy / N1 - ny * n1x / (N1 * N)
return wracc
if __name__ == "__main__":
filename = "..\\..\\doc\\datasets\\lenses.tab"
if 'linux' in sys.platform:
filename = "/usr/doc/orange/datasets/lenses.tab"
data = orange.ExampleTable(filename)
print
print
print
learner = CN2_SD(3)
targetClass = orange.Value(data.domain["lenses"], "none")
rules = learner(data, targetClass, 10)
print "____________________SN2-SD results______________________"
rules.printRules()
def makeTable(data, engine, tagLayer, useFarAway=False):
table = orange.ExampleTable(engine.domain())
if engine.name() in skipMap:
skipList = skipMap[engine.name()]
else:
skipList = []
labeledCount = 0
negativeCount = 0
farAwayCount = 0
for engineName, landmarkName, geometry in data:
#geometry["landmark"] = createLandmarkPt(math2d.centroid(geometry["landmark"]))
if not (engineName in skipList): # or True:
try:
geometry["landmark"] = geometry["ground"]
ex = engine.makeExample(**geometry)
except preposition.InsaneExample:
continue
except:
print "dc", engineName
print "dc", geometry
print "dc", landmarkName
raise
if engineName == engine.name():
if engineName == "down" and False:
print "doing down differently"
if landmarkName == "hallway":
cls = "True"
labeledCount += 1
else:
continue
#cls = "False"
#negativeCount -= 1
else:
cls = "True"
labeledCount += 1
else:
cls = "False"
negativeCount += 1
ex['class'] = cls
ex['sourceEngineName'] = engineName
ex['engineName'] = engine.name()
ex['landmarkName'] = landmarkName
ex['farAway'] = False
table.append(ex)
if useFarAway and engine.name() != "through" and engine.name() != "down":
for name, landmark in tagLayer:
centroid1 = math2d.centroid(landmark)
for engineName, landmarkName, geometry in data:
if engineName == engine.name():
centroid2 = math2d.centroid(geometry["landmark"])
d1 = math2d.dist(centroid1, centroid2)
d2 = math2d.length(geometry["figure"])
if d1 > d2:
ex = engine.makeExample(figure=geometry["figure"],
landmark=landmark)
ex['class'] = "False"
ex['landmarkName'] = landmarkName
ex['sourceEngineName'] = engineName
ex['engineName'] = engine.name()
ex['farAway'] = True
table.append(ex)
farAwayCount += 1
if farAwayCount >= 100:
break
for ex in table:
ex['drawMap'] = None
#ex['geometry'] = None
print "counts"
print labeledCount, "labeled examples."
print negativeCount, "negative examples."
print farAwayCount, "far away examples."
return table
# Classes: ExampleTable
# Uses: iris, heart_disease
# Referenced: ExampleTable.htm
import orange
import orange, random
random.seed(0)
values = ["0", "1"]
mynames = ["orange", "green", "red", "yellow", "black", "magenta"]
attributes = [orange.EnumVariable(mynames[i], values=values) for i in range(6)]
classattr = orange.EnumVariable("classname", values=["0", "1"])
domain = orange.Domain(attributes + [classattr])
print "attributes", attributes
print "classattr", classattr
print "domain:", domain
card = [1, 1, 1, 1, 1, 1]
data = orange.ExampleTable(domain)
for i in range(5):
ex = [random.randint(0, c) for c in card]
ex.append(ex[0] == ex[1] or ex[4] == 0)
data.append(ex)
for ex in data:
print ex
classifier = orange.BayesLearner(data)
print classifier(data[0], orange.GetBoth)
# Description: Add a new attribute price to a car data set, compute it from two existing attributes (buying, maint)
# Category: preprocessing
# Uses: car
# Classes: Domain, Value, getValueFrom, EnumVariable
# Referenced: domain.htm
import orange
data = orange.ExampleTable('../datasets/car')
# add attribute price = f(buying, maint)
# see also http://www.ailab.si/hint/car_dataset.asp
priceTable = {}
priceTable['v-high:v-high'] = 'v-high'
priceTable['high:v-high'] = 'v-high'
priceTable['med:v-high'] = 'high'
priceTable['low:v-high'] = 'high'
priceTable['v-high:high'] = 'v-high'
priceTable['high:high'] = 'high'
priceTable['med:high'] = 'high'
priceTable['low:high'] = 'med'
priceTable['v-high:med'] = 'high'
priceTable['high:med'] = 'high'
priceTable['med:med'] = 'med'
priceTable['low:med'] = 'low'
priceTable['v-high:low'] = 'high'
priceTable['high:low'] = 'high'
priceTable['med:low'] = 'low'
priceTable['low:low'] = 'low'
# Description: Shows how to sample example by random divisions into two groups
# Category: sampling
# Classes: MakeRandomIndices, MakeRandomIndices2, RandomGenerator
# Uses: lenses
# Referenced: RandomIndices.htm
import orange
data = orange.ExampleTable("lenses")
indices2 = orange.MakeRandomIndices2(p0=6)
ind = indices2(data)
print ind
data0 = data.select(ind, 0)
data1 = data.select(ind, 1)
print len(data0), len(data1)
print "\nIndices without playing with random generator"
for i in range(5):
print indices2(data)
print "\nIndices with random generator"
indices2.randomGenerator = orange.RandomGenerator(42)
for i in range(5):
print indices2(data)
print "\nIndices with randseed"
indices2.randomGenerator = None
indices2.randseed = 42
for i in range(5):
# Description: Learn decision tree from data and output class probabilities for first few instances
# Category: modelling
# Uses: voting.tab
# Classes: orngTree.TreeLearner
# Referenced: c_otherclass.htm
import orange, orngTree
data = orange.ExampleTable("voting")
tree = orngTree.TreeLearner(data, sameMajorityPruning=1, mForPruning=2)
print "Possible classes:", data.domain.classVar.values
print "Probabilities for democrats:"
for i in range(5):
p = tree(data[i], orange.GetProbabilities)
print "%d: %5.3f (originally %s)" % (i + 1, p[1], data[i].getclass())
print
orngTree.printTxt(tree)
orngTree.printDot(tree,
fileName='tree.dot',
internalNodeShape="ellipse",
leafShape="box")
supps = [rule.support for rule in self.rules]
self.supp_min = min(supps)
self.supp_max = max(supps)
del supps
confs = [rule.confidence for rule in self.rules]
self.conf_min = min(confs)
self.conf_max = max(confs)
del confs
self.checkScale()
else:
self.supp_min, self.supp_max = self.conf_min, self.conf_max = 0., 1.
self.supp_allmin, self.supp_allmax, self.conf_allmin, self.conf_allmax = self.supp_min, self.supp_max, self.conf_min, self.conf_max
self.rezoom(self.supp_allmin, self.supp_allmax, self.conf_allmin, self.conf_allmax)
if __name__=="__main__":
a=QApplication(sys.argv)
ow=OWAssociationRulesViewer()
dataset = orange.ExampleTable('../../doc/datasets/car.tab')
rules=orange.AssociationRulesInducer(dataset, minSupport = 0.3, maxItemSets=15000)
ow.arules(rules)
ow.show()
a.exec_()
ow.saveSettings()
# Description: Shows what the contingency matrix looks like and which are its common methods
# Category: statistics
# Classes: Contingency, ContingencyAttrClass
# Uses: monk1
# Referenced: contingency.htm
import orange
data = orange.ExampleTable("monk1")
cont = orange.ContingencyAttrClass("e", data)
print "Contingency items:"
for val, dist in cont.items():
print val, dist
print
print "Contingency keys: ", cont.keys()
print "Contingency values: ", cont.values()
print "Contingency items: ", cont.items()
print
print "cont[0] =", cont[0]
print 'cont[\"1\"] =', cont["1"]
print 'cont[orange.Value(data.domain["e"], "1")] =', cont[orange.Value(
data.domain["e"], "1")]
print
print "Iteration through contingency:"
for i in cont:
print i
print
def getSelectionsAsExampleTables(self,
attrList,
useAnchorData=1,
addProjectedPositions=0):
return (None, None) # TODO: this is disabled for now
if not self.have_data:
return (None, None)
selected = self.get_selected_indices()
if addProjectedPositions == 0 and not numpy.any(selected):
return (None, self.raw_data)
if (useAnchorData and len(self.anchor_data) < 3) or len(attrList) < 3:
return (None, None)
x_attr = orange.FloatVariable("X Positions")
y_attr = orange.FloatVariable("Y Positions")
z_attr = orange.FloatVariable("Z Positions")
if addProjectedPositions == 1:
domain = orange.Domain([x_attr, y_attr, z_attr] +
[v for v in self.data_domain.variables])
elif addProjectedPositions == 2:
domain = orange.Domain(self.data_domain)
domain.addmeta(orange.newmetaid(), x_attr)
domain.addmeta(orange.newmetaid(), y_attr)
domain.addmeta(orange.newmetaid(), z_attr)
else:
domain = orange.Domain(self.data_domain)
domain.addmetas(self.data_domain.getmetas())
if useAnchorData:
indices = [
self.attribute_name_index[val[3]] for val in self.anchor_data
]
else:
indices = [self.attribute_name_index[label] for label in attrList]
valid_data = self.getValidList(indices)
if len(valid_data) == 0:
return (None, None)
array = self.create_projection_as_numeric_array(
attrList,
scaleFactor=self.scaleFactor,
useAnchorData=useAnchorData,
removeMissingData=0)
if array == None:
return (None, None)
unselected = numpy.logical_not(selected)
selected_indices, unselected_indices = list(selected), list(unselected)
if addProjectedPositions:
selected = orange.ExampleTable(
domain, self.raw_data.selectref(selected_indices))
unselected = orange.ExampleTable(
domain, self.raw_data.selectref(unselected_indices))
selected_index = 0
unselected_index = 0
for i in range(len(selected_indices)):
if selected_indices[i]:
selected[selected_index][x_attr] = array[i][0]
selected[selected_index][y_attr] = array[i][1]
selected[selected_index][z_attr] = array[i][2]
selected_index += 1
else:
unselected[unselected_index][x_attr] = array[i][0]
unselected[unselected_index][y_attr] = array[i][1]
unselected[unselected_index][z_attr] = array[i][2]
else:
selected = self.raw_data.selectref(selected_indices)
unselected = self.raw_data.selectref(unselected_indices)
if len(selected) == 0:
selected = None
if len(unselected) == 0:
unselected = None
return (selected, unselected)
import orange
data = orange.ExampleTable("inquisition")
rules = orange.AssociationRulesSparseInducer(data,
support = 0.5, storeExamples = True)
rule0 = rules[10]
print "Rule:", rule0
print "Match left: "
print [rule0.examples[i] for i in rule0.matchLeft]
print "\nMatch both: "
print [rule0.examples[i] for i in rule0.matchBoth]
inducer = orange.AssociationRulesSparseInducer(support = 0.5)
itemsets = inducer.getItemsets(data)
print itemsets[5]
import orange
import orngCN2
data = orange.ExampleTable("titanic.tab")
# create learner
learner = orange.RuleLearner()
cl = learner(data)
for r in cl.rules:
print orngCN2.ruleToString(r)
print "*****"
learner.ruleFinder = orange.RuleBeamFinder()
learner.ruleFinder.evaluator = orngCN2.mEstimate(m=50)
cl = learner(data)
for r in cl.rules:
print orngCN2.ruleToString(r)
print "****"
learner.ruleFinder.ruleStoppingValidator = orange.RuleValidator_LRS(
alpha=0.01, min_coverage=10, max_rule_complexity=2)
learner.ruleFinder.ruleFilter = orange.RuleBeamFilter_Width(width=50)
cl = learner(data)
for r in cl.rules:
print orngCN2.ruleToString(r)
def dataset(self, data):
#self.data=data
self.data = self.isDataWithClass(
data, orange.VarTypes.Discrete) and data or None
self.setLearner()
def qualityButtonPressed(self, id=0):
self.QualityButton = id
for i in range(len(self.ruleQualityBG.buttons)):
self.ruleQualityBG.buttons[i].setChecked(id == i)
self.mSpin.control.setEnabled(id == 1)
def coveringAlgButtonPressed(self, id=0):
self.CoveringButton = id
for i in range(len(self.coveringAlgBG.buttons)):
self.coveringAlgBG.buttons[i].setChecked(id == i)
self.weightSpin.control.setEnabled(id == 1)
def applySettings(self):
self.setLearner()
if __name__ == "__main__":
app = QApplication(sys.argv)
w = OWCN2()
#w.dataset(orange.ExampleTable("titanic.tab"))
w.dataset(
orange.ExampleTable(r"E:\Development\Orange Datasets\UCI\titanic.tab"))
w.show()
app.exec_()
w.saveSettings()
def nway():
engine_to_examples = {}
trainer = Trainer()
classes = set()
for i, key in enumerate(trainer.annotationEngines):
engine = trainer.engineMap[key]
table = trainer.makeTable(engine)
for ex in table:
if ex["farAway"].value:
cls = "null"
else:
cls = ex["sourceEngineName"].value
geometry = ex["geometry"].value
engine_to_examples.setdefault(cls, [])
classes.add(cls)
examples = [
trainer.engineMap[key].makeExample(expectInsane=True,
**geometry)
for key in trainer.annotationEngines
if not len(geometry["figure"]) == 0
]
engine_to_examples[cls].append(examples)
if i >= 1:
#break
pass
variables = []
for ex in examples:
for attr in ex.domain:
if attr.name == "class":
continue
new_attr = orange.FloatVariable(attr.name)
variables.append(new_attr)
domain = orange.Domain(variables,
orange.EnumVariable("class", values=list(classes)))
table = orange.ExampleTable(domain)
for engine_name, example_lists in engine_to_examples.iteritems():
for example_list in example_lists:
ex = orange.Example(domain)
for engine_ex in example_list:
for attr in engine_ex.domain:
ex[attr.name] = engine_ex[attr.name]
ex["class"] = engine_name
table.append(ex)
print "domain", domain
cv_indices = orange.MakeRandomIndices2(table, p0=0.75)
training = table.select(cv_indices, 0, negate=True)
testing = table.select(cv_indices, 0, negate=False)
#classifier = orngBayes.BayesLearner(training)
classifier = orangePickle.PickleableClassifier(training,
orngBayes.BayesLearner)
results = orngTest.testOnData([classifier], testing)
print orngStat.CA(results)
cm = orngStat.confusionMatrices(results)[0]
classes = list(domain.classVar.values)
print " ", " ".join([c.rjust(12) for c in classes + ["", ""]])
for className, classConfusions in zip(classes, cm):
#format = ("%s" + ("\t%i" * len(classes)))
values = (className, ) + tuple(classConfusions)
print " ".join([str(c).rjust(12) for c in values])
#print format % values
for name in classes:
classIndex = classes.index(name)
mpl.figure()
rocCurve(results,
"",
classIndex,
stepSize=0.001,
plotArgs=dict(linewidth=5, markersize=10))
mpl.title(name, size=30)
mpl.xlabel("FP", fontsize=30)
mpl.ylabel("TP", fontsize=30)
mpl.xticks([0, 1], fontsize=17)
mpl.yticks([0, 1], fontsize=17)
fname = "nway.pck"
print "saving", fname
with open(fname, "w") as f:
pickle.dump(classifier, f, protocol=2)
mpl.show()
def setLearner(self):
if hasattr(self, "btnApply"):
self.btnApply.setFocus()
#progress bar
self.progressBarInit()
#learner
self.learner = orngCN2.CN2UnorderedLearner()
self.learner.name = self.name
self.learner.progressCallback = CN2ProgressBar(self)
self.send("Learner", self.learner)
ruleFinder = orange.RuleBeamFinder()
if self.QualityButton == 0:
ruleFinder.evaluator = orange.RuleEvaluator_Laplace()
elif self.QualityButton == 1:
ruleFinder.evaluator = orngCN2.mEstimate(self.m)
elif self.QualityButton == 2:
ruleFinder.evaluator = orngCN2.WRACCEvaluator()
if self.useMaxRuleLength:
maxRuleLength = self.MaxRuleLength
else:
maxRuleLength = -1
ruleFinder.ruleStoppingValidator = orange.RuleValidator_LRS(
alpha=self.stepAlpha,
min_coverage=self.MinCoverage,
max_rule_complexity=maxRuleLength)
ruleFinder.validator = orange.RuleValidator_LRS(
alpha=self.Alpha,
min_coverage=self.MinCoverage,
max_rule_complexity=maxRuleLength)
ruleFinder.ruleFilter = orange.RuleBeamFilter_Width(
width=self.BeamWidth)
self.learner.ruleFinder = ruleFinder
if self.CoveringButton == 0:
self.learner.coverAndRemove = orange.RuleCovererAndRemover_Default(
)
elif self.CoveringButton == 1:
self.learner.coverAndRemove = orngCN2.CovererAndRemover_multWeights(
mult=self.Weight)
self.classifier = None
self.error()
if self.data:
oldDomain = orange.Domain(self.data.domain)
learnData = orange.ExampleTable(oldDomain, self.data)
self.classifier = self.learner(learnData)
self.classifier.name = self.name
for r in self.classifier.rules:
r.examples = orange.ExampleTable(oldDomain, r.examples)
self.classifier.examples = orange.ExampleTable(
oldDomain, self.classifier.examples)
self.classifier.setattr("data", self.classifier.examples)
self.error("")
## except orange.KernelException, (errValue):
## self.classifier=None
## self.error(errValue)
## except Exception:
## self.classifier=None
## if not self.data.domain.classVar:
## self.error("Classless domain.")
## elif self.data.domain.classVar.varType == orange.VarTypes.Continuous:
## self.error("CN2 can learn only from discrete class.")
## else:
## self.error("Unknown error")
self.send("Classifier", self.classifier)
self.send("Unordered CN2 Classifier", self.classifier)
self.progressBarFinished()
labels = [m.name for m in d.domain.getmetas().values()] + \
[a.name for a in d.domain.variables]
self.labelCombo.addItems(labels)
# here we would need to use the domain dependent setting of the label id
self.labelCombo.setCurrentIndex(0)
self.Label = labels[0]
self.setLabel()
def dataset(self, data):
if data and len(data.domain.attributes):
self.data = data
self.setLabelComboItems()
self.computeMatrix()
else:
self.send("Distance Matrix", None)
##################################################################################################
# test script
if __name__ == "__main__":
import os
data = orange.ExampleTable(r'../../doc/datasets/glass')
data = orange.ExampleTable('glass')
a = QApplication(sys.argv)
ow = OWExampleDistance()
ow.show()
ow.dataset(data)
a.exec_()
ow.saveSettings()
# Description: Shows how to derive a Python class from orange.TreeSplitConstructor
# Category: classification, decision trees, callbacks to Python
# Classes: TreeSplitConstructor, Classifier, SubsetsGenerator_constSize, orngMisc.BestOnTheFly
# Uses: lenses
# Referenced: callbacks.htm
import orange, orngTree, orngMisc
tab = orange.ExampleTable(r"lenses.tab")
class CartesianClassifier(orange.Classifier):
def __init__(self, var1, var2):
self.var1 = var1
self.var2 = var2
self.noValues2 = len(var2.values)
self.classVar = orange.EnumVariable("%sx%s" % (var1.name, var2.name))
self.classVar.values = [
"%s-%s" % (v1, v2) for v1 in var1.values for v2 in var2.values
]
def __call__(self, ex, what=orange.Classifier.GetValue):
val = ex[self.var1] * self.noValues2 + ex[self.var2]
if what == orange.Classifier.GetValue:
return orange.Value(self.classVar, val)
probs = orange.DiscDistribution(self.classVar)
probs[val] = 1.0
if what == orange.Classifier.GetProbabilities:
return probs
else:
return (orange.Value(self.classVar, val), probs)
def setLearner(self, learner=None):
self.learner = learner
self.commit()
def setData(self, data):
self.data = data
self.commit()
def onChange(self):
pass
def commit(self):
wrapped = None
if self.learner:
wrapped = self.METHODS[self.method][1](self.learner, t=self.t)
self.send("Learner", wrapped)
if self.data and wrapped:
classifier = wrapped(self.data)
self.send("Classifier", classifier)
if __name__ == "__main__":
app = QApplication(sys.argv)
w = OWEnsemble()
w.setLearner(orange.BayesLearner())
w.setData(orange.ExampleTable("../../doc/datasets/iris"))
w.show()
app.exec_()
def __call__(self, examples, weightID=0, **kwds):
import orngTest, orngStat, statc
self.__dict__.update(kwds)
if self.removeThreshold < self.addThreshold:
raise "'removeThreshold' should be larger or equal to 'addThreshold'"
classVar = examples.domain.classVar
indices = orange.MakeRandomIndicesCV(examples,
folds=getattr(self, "folds", 10))
domain = orange.Domain([], classVar)
res = orngTest.testWithIndices([self.learner],
orange.ExampleTable(domain, examples),
indices)
oldStat = self.stat(res)[0]
oldStats = [self.stat(x)[0] for x in orngStat.splitByIterations(res)]
print(".", oldStat, domain)
stop = False
while not stop:
stop = True
if len(domain.attributes) >= 2:
bestStat = None
for attr in domain.attributes:
newdomain = orange.Domain(
[x for x in domain.attributes if x != attr], classVar)
res = orngTest.testWithIndices(
[self.learner],
(orange.ExampleTable(newdomain, examples), weightID),
indices)
newStat = self.stat(res)[0]
newStats = [
self.stat(x)[0]
for x in orngStat.splitByIterations(res)
]
print("-", newStat, newdomain)
## If stat has increased (ie newStat is better than bestStat)
if not bestStat or cmp(newStat, bestStat) == self.statsign:
if cmp(newStat, oldStat) == self.statsign:
bestStat, bestStats, bestAttr = newStat, newStats, attr
elif statc.wilcoxont(
oldStats, newStats)[1] > self.removeThreshold:
bestStat, bestAttr, bestStats = newStat, newStats, attr
if bestStat:
domain = orange.Domain(
[x for x in domain.attributes if x != bestAttr],
classVar)
oldStat, oldStats = bestStat, bestStats
stop = False
print("removed", bestAttr.name)
bestStat, bestAttr = oldStat, None
for attr in examples.domain.attributes:
if not attr in domain.attributes:
newdomain = orange.Domain(domain.attributes + [attr],
classVar)
res = orngTest.testWithIndices(
[self.learner],
(orange.ExampleTable(newdomain, examples), weightID),
indices)
newStat = self.stat(res)[0]
newStats = [
self.stat(x)[0]
for x in orngStat.splitByIterations(res)
]
print("+", newStat, newdomain)
## If stat has increased (ie newStat is better than bestStat)
if cmp(newStat,
bestStat) == self.statsign and statc.wilcoxont(
oldStats, newStats)[1] < self.addThreshold:
bestStat, bestStats, bestAttr = newStat, newStats, attr
if bestAttr:
domain = orange.Domain(domain.attributes + [bestAttr],
classVar)
oldStat, oldStats = bestStat, bestStats
stop = False
print("added", bestAttr.name)
return self.learner(orange.ExampleTable(domain, examples), weightID)
# Description: Shows different uses of orange.Domain
# Category: preprocessing
# Uses: glass
# Classes: Domain
# Referenced: domain.htm
import orange
domain = orange.ExampleTable("glass").domain
tests = (
'(["Na", "Mg"], domain)', '(["Na", "Mg"], 1, domain)',
'(["Na", "Mg"], 0, domain)', '(["Na", "Mg"], domain.variables)',
'(["Na", "Mg"], 1, domain.variables)',
'(["Na", "Mg"], 0, domain.variables)',
'([domain["Na"], "Mg"], 0, domain.variables)',
'([domain["Na"], "Mg"], 0, domain)',
'([domain["Na"], "Mg"], 0, domain.variables)',
'([domain["Na"], domain["Mg"]], 0)', '([domain["Na"], domain["Mg"]], 1)',
'([domain["Na"], domain["Mg"]], None)',
'([domain["Na"], domain["Mg"]], orange.EnumVariable("something completely different"))',
'(domain)', '(domain, 0)', '(domain, 1)', '(domain, "Mg")',
'(domain, domain[0])', '(domain, None)',
'(domain, orange.FloatVariable("nothing completely different"))')
for args in tests:
line = "orange.Domain%s" % args
d = eval(line)
print line
print " classVar: %s" % d.classVar
print " attributes: %s" % d.attributes
import orange
import orngClustering
def callback(km):
print "Iteration: %d, changes: %d, score: %.4f" % (km.iteration,
km.nchanges, km.score)
data = orange.ExampleTable("iris")
km = orngClustering.KMeans(data, 3, minscorechange=0, inner_callback=callback)
# Description: Builds a regression tree and prints it out
# Category: modelling
# Uses: housing
# Classes: orngTree.TreeLearner
# Referenced: regression.htm
import orange, orngTree
data = orange.ExampleTable("housing")
rt = orngTree.TreeLearner(data, measure="retis", mForPruning=2, minExamples=20)
orngTree.printTxt(rt, leafStr="%V %I")
# Description: Read data, show mean for continuous attributes and contingency matrix for nominal attributes
# Category: description
# Uses: adult_sample.tab
# Classes: DomainContingency
# Referenced: basic_exploration.htm
import orange
data = orange.ExampleTable("../datasets/adult_sample")
print "Continuous attributes:"
for a in range(len(data.domain.attributes)):
if data.domain.attributes[a].varType == orange.VarTypes.Continuous:
d = 0.
n = 0
for e in data:
if not e[a].isSpecial():
d += e[a]
n += 1
print " %s, mean=%3.2f" % (data.domain.attributes[a].name, d / n)
print "\nNominal attributes (contingency matrix for classes:", data.domain.classVar.values, ")"
cont = orange.DomainContingency(data)
for a in data.domain.attributes:
if a.varType == orange.VarTypes.Discrete:
print " %s:" % a.name
for v in range(len(a.values)):
sum = 0
for cv in cont[a][v]:
sum += cv
print " %s, total %d, %s" % (a.values[v], sum, cont[a][v])
print
def __call__(self, data, weight=None):
if not self.use_attributes == None:
new_domain = orange.Domain(self.use_attributes,
data.domain.classVar)
new_domain.addmetas(data.domain.getmetas())
data = orange.ExampleTable(new_domain, data)
if self.stepwise and self.stepwise_before:
use_attributes = stepwise(data,
add_sig=self.add_sig,
remove_sig=self.remove_sig)
new_domain = orange.Domain(use_attributes, data.domain.classVar)
new_domain.addmetas(data.domain.getmetas())
data = orange.ExampleTable(new_domain, data)
# continuization (replaces discrete with continuous attributes)
continuizer = orange.DomainContinuizer()
continuizer.multinomialTreatment = continuizer.FrequentIsBase
continuizer.zeroBased = True
domain0 = continuizer(data)
data = data.translate(domain0)
if self.stepwise and not self.stepwise_before:
use_attributes = stepwise(data,
weight,
add_sig=self.add_sig,
remove_sig=self.remove_sig)
new_domain = orange.Domain(use_attributes, data.domain.classVar)
new_domain.addmetas(data.domain.getmetas())
data = orange.ExampleTable(new_domain, data)
# missing values handling (impute missing)
imputer = orange.ImputerConstructor_model()
imputer.learnerContinuous = orange.MajorityLearner()
imputer.learnerDiscrete = orange.MajorityLearner()
imputer = imputer(data)
data = imputer(data)
# convertion to numpy
A, y, w = data.toNumpy() # weights ??
if A == None:
n = len(data)
m = 0
else:
n, m = numpy.shape(A)
if self.beta0 == True:
if A == None:
X = numpy.ones([len(data), 1])
else:
X = numpy.insert(A, 0, 1, axis=1) # adds a column of ones
else:
X = A
# set weights
W = numpy.identity(len(data))
if weight:
for di, d in enumerate(data):
W[di, di] = float(d[weight])
D = dot(
dot(numpy.linalg.pinv(dot(dot(X.T, W), X)), X.T), W
) # adds some robustness by computing the pseudo inverse; normal inverse could fail due to singularity of the X.T*W*X
beta = dot(D, y)
yEstimated = dot(X, beta) # estimation
# some desriptive statistisc
muY, sigmaY = numpy.mean(y), numpy.std(y)
muX, covX = numpy.mean(X, axis=0), numpy.cov(X, rowvar=0)
# model statistics
SST, SSR = numpy.sum((y - muY)**2), numpy.sum((yEstimated - muY)**2)
SSE, RSquare = SST - SSR, SSR / SST
R = numpy.sqrt(RSquare) # coefficient of determination
RAdjusted = 1 - (1 - RSquare) * (n - 1) / (n - m - 1)
F = (SSR / m) / (SST - SSR / (n - m - 1)) # F statistisc
df = m - 1
sigmaSquare = SSE / (n - m - 1)
# standard error of estimated coefficients
errCoeff = sqrt(sigmaSquare * inv(dot(X.T, X)).diagonal())
# t statistisc, significance
t = beta / errCoeff
df = n - 2
significance = []
for tt in t:
try:
significance.append(
statc.betai(df * 0.5, 0.5, df / (df + tt * tt)))
except:
significance.append(1.0)
# standardized coefficients
if m > 0:
stdCoeff = (sqrt(covX.diagonal()) / sigmaY) * beta
else:
stdCoeff = (sqrt(covX) / sigmaY) * beta
model = {
'descriptives': {
'meanX': muX,
'covX': covX,
'meanY': muY,
'sigmaY': sigmaY
},
'model': {
'estCoeff': beta,
'stdErrorEstimation': errCoeff
},
'model summary': {
'TotalVar': SST,
'ExplVar': SSE,
'ResVar': SSR,
'R': R,
'RAdjusted': RAdjusted,
'F': F,
't': t,
'sig': significance
}
}
return LinearRegression(statistics=model,
domain=data.domain,
name=self.name,
beta0=self.beta0,
imputer=imputer)
print "Attributes in favor of %s = %s [%f]" % (
t.domain.classVar.name, t.domain.classVar.values[1],
m.probfunc(m.example_c[idx][0]))
printpie(e1, m.probfunc(m.example_c[idx][0]))
print "\nProjection of the example in the basis space:"
j = 0
for i in range(len(m.coeff_names)):
print m.coeff_names[i][0], ':'
for x in m.coeff_names[i][1:]:
print '\t', x, '=', vector[j]
j += 1
print "beta:", -m.beta
#t = orange.ExampleTable('c:/proj/domains/voting.tab') # discrete
t = orange.ExampleTable(
r"E:\Development\Orange Datasets\UCI\shuttle.tab") # discrete
#t = orange.ExampleTable('c_cmc.tab') # continuous
print "NAIVE BAYES"
print "==========="
bl = orange.BayesLearner()
bl.estimatorConstructor = orange.ProbabilityEstimatorConstructor_Laplace()
# prevent too many estimation points
# increase the smoothing level
bl.conditionalEstimatorConstructorContinuous = orange.ConditionalProbabilityEstimatorConstructor_loess(
windowProportion=0.5, nPoints=10)
c = bl(t)
printmodel(t, c, printexamples=0)
print "\n\nLOGISTIC REGRESSION"
## data = orange.ExampleTable(r'..\..\doc\datasets\adult_sample.tab')
## dataA = orange.ExampleTable(r'c:\Documents and Settings\peterjuv\My Documents\STEROLTALK\Sterolgene v.0 mouse\sterolgene v.0 mouse probeRatios.tab')
## dataA = orange.ExampleTable(r'c:\Documents and Settings\peterjuv\My Documents\STEROLTALK\Sterolgene v.0 mouse\Copy of sterolgene v.0 mouse probeRatios.tab')
## dataB = orange.ExampleTable(r'c:\Documents and Settings\peterjuv\My Documents\STEROLTALK\Sterolgene v.0 mouse\sterolgene v.0 mouse probeRatios.tab')
dataA = orange.ExampleTable(r'c:\Documents and Settings\peterjuv\My Documents\et1.tab')
dataB = orange.ExampleTable(r'c:\Documents and Settings\peterjuv\My Documents\et2.tab')
a=QApplication(sys.argv)
ow=OWMergeData()
a.setMainWidget(ow)
ow.show()
ow.onDataAInput(dataA)
ow.onDataBInput(dataB)
# data table
dt = OWDataTable.OWDataTable(signalManager = signalManager)
signalManager.addWidget(ow)
signalManager.addWidget(dt)
signalManager.setFreeze(1)
signalManager.addLink(ow, dt, 'Merged Examples A+B', 'Examples', 1)
signalManager.addLink(ow, dt, 'Merged Examples B+A', 'Examples', 1)
signalManager.setFreeze(0)
dt.show()
a.exec_()
"""
import sys
a = QApplication(sys.argv)
ow = OWMergeData()
ow.show()
data = orange.ExampleTable("iris.tab")
ow.onDataAInput(data)
a.exec_()
# Description: using your own imputer and continuizer in PCA
# Category: projection
# Uses: adult_sample
# Referenced: orngPCA.htm
# Classes: orngPCA.PCA
import orange, orngPCA
data = orange.ExampleTable("bridges.tab")
imputer = orange.ImputerConstructor_maximal
continuizer = orange.DomainContinuizer()
continuizer.multinomialTreatment = continuizer.AsNormalizedOrdinal
continuizer.classTreatment = continuizer.Ignore
continuizer.continuousTreatment = continuizer.Leave
pca = orngPCA.PCA(data,
standardize=True,
imputer=imputer,
continuizer=continuizer)
print pca