def myDist(vecA, vecB): distSpotA = latlonDist._dist( (vecA.item(0,0),vecA.item(0,1)) ,(vecA.item(0,3), vecA.item(0,4)) ) distSpotB = latlonDist._dist( (vecB.item(0,0),vecB.item(0,1)) ,(vecB.item(0,3), vecB.item(0,4)) ) # print distSpotB # print distSpotA distSpotAB = abs(distSpotA - distSpotB) distAB = latlonDist._dist( (vecA.item(0,3),vecA.item(0,4)) ,(vecB.item(0,3),vecB.item(0,4)) ) diffNotesAB = abs( vecA.item(0,2)-vecB.item(0,2) ) dist = distSpotAB+distAB+diffNotesAB return dist
def myDist(vecA, vecB):
distSpotA = latlonDist._dist((vecA.item(0, 0), vecA.item(0, 1)),
(vecA.item(0, 3), vecA.item(0, 4)))
distSpotB = latlonDist._dist((vecB.item(0, 0), vecB.item(0, 1)),
(vecB.item(0, 3), vecB.item(0, 4)))
# print distSpotB
# print distSpotA
distSpotAB = abs(distSpotA - distSpotB)
distAB = latlonDist._dist((vecA.item(0, 3), vecA.item(0, 4)),
(vecB.item(0, 3), vecB.item(0, 4)))
diffNotesAB = abs(vecA.item(0, 2) - vecB.item(0, 2))
dist = distSpotAB + distAB + diffNotesAB
return dist
def genAll():
clusterLabels = csv.reader(open("clustersLabels.csv","r"))
touristInfo = csv.reader(open("touristInfo.csv","r"))
dictAll = {}
touristDict = {}
for row in touristInfo:
touristDict[row[0]] = [row[1],row[2],row[3]]
count = 0
for row in clusterLabels:
if row[0] in touristDict:
for item in touristDict[row[0]]:
row.append(item)
dictAll[count] = row
count+=1
for key, value in dictAll.iteritems():
temp = []
spot = (float(value[1]),float(value[2]))
residence = (float(value[-2]), float(value[-1]))
dist = latlonDist._dist(residence,spot)
temp.append(value[0]) #uid
temp.append(value[3]) #clusterLabel
temp.append(value[5]) #noteNumber
temp.append(dist) #distResidenceSpot
temp.append(value[4]) #publicationDate
dictAll[key] = temp
return dictAll
def convertGeoCode(filename,destname):
geocodeBuffer = genGeoCodeDict()
writeRows = []
with open(filename) as rawCSV:
rawContent = csv.reader(rawCSV)
for row in rawContent:
if row[1] != 'null' and row[1] in geocodeBuffer:
distance = latlonDist._dist(geocodeBuffer[row[1]],geocodeBuffer[destname])
writeRows.append((0,row[2],distance,0))
print '%d tours are found' % len(writeRows)
with open('data/'+filename,'wb') as destCSV:
writer = csv.writer(destCSV)
for row in writeRows:
writer.writerow(row)
def _dist(p,q):
assert type(p) is np.matrix, 'p is not matrix!'
assert type(q) is np.matrix, 'q is not matrix!'
spot1 = (p.item(0,1), p.item(0,0))
spot2 = (q.item(0,1), q.item(0,0))
return latlonDist._dist(spot1,spot2)
def getDistance(p, q):
assert type(p) is np.ndarray, 'p is not matrix!'
assert type(q) is np.ndarray, 'q is not matrix!'
pointA = (p[1], p[0])
pointB = (q[1], q[0])
return latlonDist._dist(pointA, pointB)
def distSpotTourist(dataSet): n = shape(dataSet)[0] # get row number of dataSet for i in range(n): dataSet[i,1] = latlonDist._dist( (dataSet[i,0],dataSet[i,1]), (dataSet[i,3],dataSet[i,4]) ) return dataSet[:,1:3]
def distSpotTourist(dataSet):
n = shape(dataSet)[0] # get row number of dataSet
for i in range(n):
dataSet[i, 1] = latlonDist._dist((dataSet[i, 0], dataSet[i, 1]),
(dataSet[i, 3], dataSet[i, 4]))
return dataSet[:, 1:3]
def _dist(p, q):
assert type(p) is np.matrix, 'p is not matrix!'
assert type(q) is np.matrix, 'q is not matrix!'
spot1 = (p.item(0, 1), p.item(0, 0))
spot2 = (q.item(0, 1), q.item(0, 0))
return latlonDist._dist(spot1, spot2)
