def burt_table(data, attributes):
""" Construct a Burt table (all values cross-tabulation) from data for attributes.
Return and ordered list of (attribute, value) pairs and a numpy.ndarray with the tabulations.
:param data: Data table.
:type data: :class:`Orange.data.Table`
:param attributes: List of attributes (must be Discrete).
:type attributes: list
Example ::
>>> data = Orange.data.Table("smokers_ct")
>>> items, counts = burt_table(data, [data.domain["Staff group"], data.domain["Smoking category"]])
"""
values = [(attr, value) for attr in attributes for value in attr.values]
table = numpy.zeros((len(values), len(values)))
counts = [len(attr.values) for attr in attributes]
offsets = [sum(counts[:i]) for i in range(len(attributes))]
for i in range(len(attributes)):
for j in range(i + 1):
attr1 = attributes[i]
attr2 = attributes[j]
cm = contingency.VarVar(attr1, attr2, data)
cm = numpy.array([list(row) for row in cm])
range1 = range(offsets[i], offsets[i] + counts[i])
range2 = range(offsets[j], offsets[j] + counts[j])
start1, end1 = offsets[i], offsets[i] + counts[i]
start2, end2 = offsets[j], offsets[j] + counts[j]
table[start1:end1, start2:end2] += cm
if i != j: #also fill the upper part
table[start2:end2, start1:end1] += cm.T
return values, table
## [95, 58, 66, 84, 41],
## [80, 73, 83, 4 , 96],
## [79, 93, 35, 73, 63]])
##
## data = [[9, 11, 4],
## [ 3, 5, 3],
## [ 11, 6, 3],
## [24, 73, 48]]
# Author punctuation (from 'Correspondence Analysis - Herve Abdi Lynne J. Williams')
data = [[7836, 13112, 6026], [53655, 102383, 42413],
[115615, 184541, 59226], [161926, 340479, 62754],
[38177, 105101, 12670], [46371, 58367, 14299]]
c = CA(
data,
["Rousseau", "Chateaubriand", "Hugo", "Zola", "Proust", "Giraudoux"],
["period", "comma", "other"])
c.plot_scree_diagram()
c.plot_biplot()
import Orange
data = Orange.data.Table("../../doc/datasets/smokers_ct")
staff = data.domain["Staff group"]
smoking = data.domain["Smoking category"]
cont = contingency.VarVar(staff, smoking, data)
c = CA(cont, staff.values, smoking.values)
c.plot_scree_diagram()
c.plot_biplot()
# Description: Demonstrates the use of correspondence analysis
# Category: correspondence, projection
# Classes: CA
# Uses: bridges.tab
import Orange
import Orange.projection.correspondence as corr
import Orange.statistics.contingency as cont
bridges = Orange.data.Table("bridges")
cm = cont.VarVar("PURPOSE", "MATERIAL", bridges)
ca = corr.CA(cm)
def report(coors, labels):
for coor, label in zip(coors, labels):
print " %-10s (%.3f, %.3f)" % (label + ":", coor[0, 0], coor[0, 1])
print "PURPOSE"
report(ca.column_factors(), bridges.domain["PURPOSE"].values)
print
print "MATERIAL"
report(ca.row_factors(), bridges.domain["PURPOSE"].values)
print
