def runCA(self):
self.contColAttr = colAttr = self.allAttrs[self.colAttr]
self.contRowAttr = rowAttr = self.allAttrs[self.rowAttr]
self.contingency = orange.ContingencyAttrAttr(rowAttr, colAttr,
self.data)
self.error(0)
try:
self.CA = orngCA.CA([[c for c in row] for row in self.contingency])
except numpy.linalg.LinAlgError:
self.error(0, "Could not compute the mapping! " + str(ex))
self.graph.removeDrawingCurves(True, True, True)
raise
self.rowItems = [
s for s, v in self.contingency.outerDistribution.items()
]
self.colItems = [
s for s, v in self.contingency.innerDistribution.items()
]
self.xAxisCB.clear()
self.yAxisCB.clear()
self.axisCount = min(self.CA.D.shape)
self.xAxisCB.addItems([str(i + 1) for i in range(self.axisCount)])
self.yAxisCB.addItems([str(i + 1) for i in range(self.axisCount)])
self.xPrincipalAxis = min(self.xPrincipalAxis, self.axisCount - 1)
self.yPrincipalAxis = min(self.yPrincipalAxis, self.axisCount - 1)
self.updateGraph()
def updateTables(self):
if self.textData:
if textCorpusModul:
data = (self.attrRow == 'document' and [self.data]
or [CategoryDocument(self.data).dataCD])[0]
else:
data = self.data
metas = data.domain.getmetas(orngText.TEXTMETAID)
lenMetas = len(metas)
caList = []
for ex in data:
cur = [0] * lenMetas
for i, m in zip(range(lenMetas), metas.keys()):
try:
cur[i] = float(ex[m].native())
except:
cur[i] = 0
caList.append(cur)
if not self.CAloaded:
self.CA = orngCA.CA(caList)
hasNameAttribute = 'name' in [
i.name for i in data.domain.attributes
]
hasCategoryAttribute = 'category' in [
i.name for i in data.domain.attributes
]
if not hasCategoryAttribute:
if not hasNameAttribute:
self.tipsR = [ex['text'].value[:35] for ex in data]
self.rowCategories = [(ex['text'].value[:35], "Row points")
for ex in data]
self.catColors = {"Row points": 0}
else:
self.tipsR = [ex['name'].native() for ex in data]
self.rowCategories = [(ex['name'].native(), "Row points")
for ex in data]
self.catColors = {"Row points": 0}
try:
if hasCategoryAttribute:
if not hasNameAttribute:
self.tipsR = [ex['text'].value[:35] for ex in data]
self.rowCategories = [(ex['text'].value[:35],
ex['category'].native())
for ex in data]
else:
self.tipsR = [ex['name'].native() for ex in data]
self.rowCategories = [(ex['name'].native(),
ex['category'].native())
for ex in data]
self.catColors = {}
col = 0
colors = [0, 2, 3, 5, 6, 12]
for ex in data:
if ex['category'].native() not in list(
self.catColors.keys()):
self.catColors[
ex['category'].native()] = colors[col]
col = (col + 1) % len(colors)
except:
if hasCategoryAttribute:
if not hasNameAttribute:
self.tipsR = [ex['text'].value[:35] for ex in data]
self.rowCategories = [(ex['text'].value[:35],
ex[-1].native()) for ex in data]
else:
self.tipsR = [ex.name for ex in data]
self.rowCategories = [(ex.name, ex[-1].native())
for ex in data]
self.catColors = {}
col = 0
colors = [0, 2, 3, 5, 6, 12]
for ex in data:
if ex['category'].native() not in list(
self.catColors.keys()):
self.catColors[
ex['category'].native()] = colors[col]
col = (col + 1) % len(colors)
self.tipsC = [
a.name
for a in data.domain.getmetas(orngText.TEXTMETAID).values()
]
else:
ca = orange.ContingencyAttrAttr(self.attrRow, self.attrCol,
self.data)
caList = [[col for col in row] for row in ca]
if not self.CAloaded:
self.CA = orngCA.CA(caList)
self.tipsR = [s for s, v in ca.outerDistribution.items()]
self.tipsC = [s for s, v in ca.innerDistribution.items()]
self.rowCategories = [(t, "Row points") for t in self.tipsR]
self.catColors = {"Row points": 0}
del ca
self.rowSlider.setMinimum(1)
self.rowSlider.setMaximum(len(self.tipsR))
self.percRow = len(self.tipsR) > 100 and 0.5 * len(self.tipsR) or len(
self.tipsR)
self.colSlider.setMinimum(1)
self.colSlider.setMaximum(len(self.tipsC))
self.percCol = len(self.tipsC) > 100 and 0.5 * len(self.tipsC) or len(
self.tipsC)
self.initAxesValues()
self.tabsMain.setCurrentWidget(self.graph)
self.calcRadius()
del caList
# Description: Demonstrates the use of correspondence analysis
# Category: correspondence, projection
# Classes: CA
# Uses: bridges.tab
import orange
import orngCA
data = orange.ExampleTable("bridges")
cm = orange.ContingencyAttrAttr("PURPOSE", "MATERIAL", data)
ca = orngCA.CA([list(col) for col in cm])
def report(coors, labels):
for coor, label in zip(coors, labels):
print " %-10s (%.3f, %.3f)" % (label + ":", coor[0], coor[1])
print "PURPOSE"
report(ca.getPrincipalColProfilesCoordinates(), data.domain["PURPOSE"].values)
print
print "MATERIAL"
report(ca.getPrincipalRowProfilesCoordinates(), data.domain["PURPOSE"].values)
print