def interregionalInequalityTestOneVariable(Y, area2region, permutations=9999):
def getVar(Y, possition):
result = {}
for k in Y:
result[k] = [Y[k][possition]]
return result
def shuffleMap(Y):
result = {}
values = Y.values()
numpy.random.shuffle(values)
keys = Y.keys()
newY = dict(zip(keys, values))
return newY
results = []
for nv1 in range(len(Y[0])):
var = getVar(Y, nv1)
t1, tb1, tw1 = theil(var, area2region)
numerator = 1
for iter in range(permutations):
var = shuffleMap(var)
t2, tb2, tw2 = theil(var, area2region)
if tb1 <= tb2:
numerator += 1
results.append(numerator / float(permutations + 1))
return results
def interregionalInequalityTestOneVariable(Y, area2region, permutations=9999):
def getVar(Y, possition):
result = {}
for k in Y:
result[k] = [Y[k][possition]]
return result
def shuffleMap(Y):
result = {}
values = Y.values()
numpy.random.shuffle(values)
keys = Y.keys()
newY = dict(zip(keys,values))
return newY
results = []
for nv1 in range(len(Y[0])):
var = getVar(Y,nv1)
t1,tb1,tw1 = theil(var,area2region)
numerator = 1
for iter in range(permutations):
var = shuffleMap(var)
t2,tb2,tw2 = theil(var,area2region)
if tb1 <= tb2:
numerator += 1
results.append(numerator/float(permutations+1))
return results
def inequalityMultivar(Y, area2region, index='theil'):
"""Inequality index for multiple variables
This function calculates a given inequality index for multiple variables::
Layer.inequality('inequalityMultivar',vars, cluster, <index>)
:keyword vars: List with variables to be analyzed; e.g: ['Y1978', 'Y1979', 'Y1980', 'Y1981']
:type vars: list
:keyword cluster: variable in Layer containing regionalization solution; e.g.: 'BELS'
:type cluster: string
:keyword index: inequality index to be applied. Default value index = 'theil'.
:type index: string
:rtype: Tuple
:return: index, index between groups, index within groups, index whitin groups over index
**Example 1** ::
import clusterpy
instance = clusterpy.importArcData("clusterpy/data_examples/china")
result = instance.inequality('inequality',['Y1978', 'Y1979', 'Y1980', 'Y1981'], 'BELS', index = 'theil')
"""
matrix = numpy.matrix(Y.values()).transpose()
periods = [x.tolist()[0] for x in matrix]
areas = Y.keys()
t = []
tb = []
tw = []
tw_t = []
for var in periods:
var = [[x] for x in var]
dictionary = dict(zip(areas, var))
if index == 'theil':
t2, tb2, tw2 = theil(dictionary, area2region)
tw_t2 = tw2 / float(t2)
else:
raise NameError("index is not available in clusterpy")
t.append(t2)
tb.append(tb2)
tw.append(tw2)
tw_t.append(tw_t2)
return t, tb, tw, tw_t
def inequalityMultivar(Y, area2region, index = 'theil'):
"""Inequality index for multiple variables
This function calculates a given inequality index for multiple variables::
Layer.inequality('inequalityMultivar',vars, cluster, <index>)
:keyword vars: List with variables to be analyzed; e.g: ['Y1978', 'Y1979', 'Y1980', 'Y1981']
:type vars: list
:keyword cluster: variable in Layer containing regionalization solution; e.g.: 'BELS'
:type cluster: string
:keyword index: inequality index to be applied. Default value index = 'theil'.
:type index: string
:rtype: Tuple
:return: index, index between groups, index within groups, index whitin groups over index
**Example 1** ::
import clusterpy
instance = clusterpy.importArcData("clusterpy/data_examples/china")
result = instance.inequality('inequality',['Y1978', 'Y1979', 'Y1980', 'Y1981'], 'BELS', index = 'theil')
"""
matrix = numpy.matrix(Y.values()).transpose()
periods = [x.tolist()[0] for x in matrix]
areas = Y.keys()
t = []
tb = []
tw = []
tw_t = []
for var in periods:
var = [[x] for x in var]
dictionary = dict(zip(areas,var))
if index == 'theil':
t2,tb2,tw2 = theil(dictionary,area2region)
tw_t2 = tw2/float(t2)
else:
raise NameError("index is not available in clusterpy")
t.append(t2)
tb.append(tb2)
tw.append(tw2)
tw_t.append(tw_t2)
return t,tb,tw,tw_t
def globalInequalityChanges(Y, fieldNames, outFile, permutations=9999):
"""Global inequality change test
This function tests whether global inequality has significantly changed
for the Theil statistic over the period t to t+k. For more information on
this function see [Rey_Sastre2010] (this function recreates Table 2 in
that paper).
Layer.inequality('globalInequalityChanges', var, outFile, <permutations>)
:keyword var: List with variables to be analyzed; e.g: ['Y1978', 'Y1979', 'Y1980', 'Y1981']
:type var: list
:keyword outFile: Name for the output file; e.g.: "regionsDifferenceTest.csv"
:type outFile: string
:keyword permutations: Number of random spatial permutations. Default value permutations = 9999.
:type permutations: integer
**Example 1** ::
import clusterpy
china = clusterpy.importArcData("clusterpy/data_examples/china")
result = china.inequality('globalInequalityChanges',['Y1978', 'Y1979', 'Y1980', 'Y1981'], "interregional_inequality_differences.csv")
"""
def getVar(Y, possition):
result = {}
for k in Y:
result[k] = [Y[k][possition]]
return result
def shufflePeriods(Y,pos1,pos2):
result = {}
for k in Y:
possibilities = [Y[k][pos1],Y[k][pos2]]
result[k] = [possibilities[numpy.random.randint(0,2)]]
return result
print "Creating global Inequality Changes [Rey_Sastre2010 - Table 2]"
results = {}
r2a = range(len(Y))
for nv1, var1 in enumerate(fieldNames):
var = getVar(Y,nv1)
t1,tb1,tw1 = theil(var,r2a)
results[(var1,var1)] = t1
for nv2, var2 in enumerate(fieldNames[nv1+1:]):
var = getVar(Y,nv1+nv2+1)
t2,tb2,tw2 = theil(var,r2a)
results[(var1,var2)] = t2 - t1
numerator = 1
for iter in range(permutations):
var = shufflePeriods(Y,nv1,nv1 + nv2 + 1)
t3,tb3,tw3 = theil(var,r2a)
if abs(t2-t1) < abs(t3-t1):
numerator += 1
results[(var2,var1)] = numerator/float(permutations+1)
if outFile:
fout = open(outFile,"w")
aux = str(fieldNames).replace("[","")
aux = aux.replace("]","")
aux = aux.replace("'","")
line = "".join([",",aux])
fout.write("".join([line,"\n"]))
for var1 in fieldNames:
line = [var1]
for var2 in fieldNames:
line += [results[(var1,var2)]]
line = str(line).replace("[","")
line = line.replace("]","")
line = line.replace("'","")
fout.write("".join([line,"\n"]))
fout.close()
print "global Inequality Changes created!"
return results
def interregionalInequalityDifferences(Y, fieldNames, area2regions,
area2regionsNames, outFile="", permutations=9999):
"""Interregional inequality differences
This function examines whether the differences across a set of clustering
solutions are significant. For more information on this function see
[Rey_Sastre2010] (this function recreates Table 6 in that paper).
Layer.inequality('interregionalInequalityDifferences', var, clusters, outFile="", <permutations>)
:keyword var: List with variables to be analyzed; e.g: ['Y1978', 'Y1979', 'Y1980', 'Y1981']
:type var: list
:keyword clusters: variables in Layer containing regionalization schemes e.g.: ["arisel1", "arisel2", "arisel3", "BELS"]
:type clusters: list
:keyword outFile: Name for the output file; e.g.: "regionsDifferenceTest.csv"
:type outFile: string
:keyword permutations: Number of random spatial permutations. Default value permutations = 9999.
:type permutations: integer
**Example 1** ::
import clusterpy
china = clusterpy.importArcData("clusterpy/data_examples/china")
china.inequality('regionsInequalityDifferenceTest',['Y1978', 'Y1979', 'Y1980', 'Y1981'], ['BELS','T78-98','T86_98'], "interregional_inequality_differences.csv")
"""
def getVar(Y, possition):
result = {}
for k in Y:
result[k] = [Y[k][possition]]
return result
def shuffleMap(Y):
values = Y.values()
numpy.random.shuffle(values)
keys = Y.keys()
newY = dict(zip(keys,values))
return newY
results = {}
for nv1, v1 in enumerate(fieldNames):
for na2r_1,a2r_1 in enumerate(area2regions):
var = getVar(Y,nv1)
t1,tb1,tw1 = theil(var,a2r_1)
name = area2regionsNames[na2r_1]
results[(v1,name,name)] = tb1
for na2r_2,a2r_2 in enumerate(area2regions[na2r_1+1:]):
t2,tb2,tw2 = theil(var,a2r_2)
oDifference = tb1 - tb2
numerator = 1
for iter in range(permutations):
var = shuffleMap(var)
t3,tb3,tw3 = theil(var,a2r_1)
t4,tb4,tw4 = theil(var,a2r_2)
rDifference = tb3 - tb4
if abs(rDifference) <= abs(oDifference):
numerator += 1
p = numerator/float(permutations+1)
name2 = area2regionsNames[na2r_2 + na2r_1 + 1]
results[(v1,name,name2)] = oDifference
results[(v1,name2,name)] = p
if outFile:
fout = open(outFile,"w")
for var1 in fieldNames:
aux = str(area2regionsNames).replace("[","")
aux = aux.replace("]","")
aux = aux.replace("'","")
line = "".join([var1,",",aux])
fout.write("".join([line,"\n"]))
for a2r in area2regionsNames:
line = [a2r]
for a2r2 in area2regionsNames:
line += [results[(var1,a2r,a2r2)]]
line = str(line).replace("[","")
line = line.replace("]","")
line = line.replace("'","")
fout.write("".join([line,"\n"]))
fout.write("\n")
fout.close()
return None
