def updateChiSquareIndependence(self, syncNumber, get):
"""Update a baseline model with a chiSquareIndependence
testStatistic.
This updates a multi-dimensional CountTable with the data,
entirely encoded in PMML.
"""
self.resetLoggerLevels()
if self.first:
if not self._updateChiSquareIndependence_first(get):
return False
self.first = False
got = {}
key = []
for f in self.fields:
value = get(f)
if value is INVALID or value is MISSING:
self.logger.debug(
"updateChiSquareIndependence: returning False (INVALID or MISSING data)"
)
return False
got[f] = value
key.append(value)
key = tuple(key)
depth, node = self._updateChiSquareIndependence_findNode(
0, got, self.countTable)
# if we don't have a table entry for this yet, make one
for i in xrange(depth, len(self.fields)):
f = self.fields[i]
if i == len(self.fields) - 1:
child = pmml.FieldValueCount(field=f, value=got[f], count=0.)
else:
child = pmml.FieldValue(field=f, value=got[f])
node.children.append(child)
node = child
updator = self.updators.get(key, None)
if updator is None:
updator = self.engine.producerUpdateScheme.updator(SUMX)
updator.initialize({SUMX: node.attrib["count"]})
self.updators[key] = updator
# for histograms, increment all bins, but only the correct bin gets a non-zero value
for otherupdator in self.updators.values():
if updator is not otherupdator:
otherupdator.increment(syncNumber, 0.)
updator.increment(syncNumber, 1.)
node.attrib["count"] = updator.sum()
self.total_updator.increment(syncNumber, 1.)
self.countTable.attrib["sample"] = self.total_updator.sum()
return True
def _updateChiSquareIndependence_newTableFromFields(self, got):
self.countTable = pmml.CountTable(sample=0.)
dimension = self.countTable
for i, f in enumerate(self.fields):
if i == len(self.fields) - 1:
dimension.children.append(pmml.FieldValueCount(field=f, value=got[f], count=0.))
else:
dimension.children.append(pmml.FieldValue(field=f, value=got[f]))
dimension = dimension.child(pmml.nonExtension)
self.baseline.children = [self.countTable]