def main():
laptops = doSearch("laptop")
print laptops[0:10]
# print getCategory('computers')
# print getCategory('laptops', parentID=58058)
# if laptops:
# print getItem(laptops[7][0])
set = makeLaptopDataset()
numpredict.knnestimate(set, (512, 1000, 14, 40, 1000))
def getkNN(data, bName, bData):
k = [1, 2, 5, 10, 20]
for x in k:
neighbors = numpredict.knnestimate(data, bData, x)
print("Nearest Neighbors For ", bName, "(k = ", x, ")")
print("Nearest Neighbors For ",
bName,
"(k = ",
x,
")",
file=open(str(bName) + ".txt", 'a+'))
for y in neighbors:
nIndex = y[1]
nName = names[nIndex]
print("\t", nName, "\t---\t", y[0])
print("\t",
nName,
"\t---\t",
y[0],
file=open(str(bName) + ".txt", 'a+'))
print("")
print("", file=open(str(bName) + ".txt", 'a+'))
def codeForKNearest(): ''' 计算输入点与训练集距离最近的k=5各点，然后将这k个price取平均值作为预测价格 ''' print '## Code for k-Nearest Neighbors 计算KNN' global data reload(numpredict) print numpredict.knnestimate(data,(95.0,3.0)) print numpredict.knnestimate(data,(99.0,3.0)) print numpredict.knnestimate(data,(99.0,5.0)) print numpredict.wineprice(99.0,5.0) # Get the actual price print numpredict.knnestimate(data,(99.0,5.0),k=1) # Try with fewer neighbors
def stemmeda8():
# behave exactly as its non-stemmed counterpart except using the stemmed file
blognames, words, data = readfile("datafiles/blogtop500_stemmed.txt")
fmeasure_idk = blognames.index('F-Measure')
wsdl_idk = blognames.index('Web Science and Digital Libraries Research Group')
takeout = data.pop(fmeasure_idk)
mout = []
klast = None
print("http://f-measure.blogspot.com/")
for k in [1, 2, 5, 10, 20]:
if k != 1:
val = knnestimate(data, data[fmeasure_idk]['input'], k)[klast:]
else:
val = knnestimate(data, data[fmeasure_idk]['input'], k)
klast = k
mout.append(["%d"%k, ", ".join(val)])
print("The %d nearest neighbors are:" % k, knnestimate(data, data[fmeasure_idk]['input'], k))
headers = ['k', 'nearest neighbors']
with open("tables/fmeasure-knearest-stemmed.text", "w+") as out:
out.write(tabulate(mout, headers=headers, tablefmt="latex"))
data.insert(fmeasure_idk, takeout)
print("http://ws-dl.blogspot.com/")
data.pop(wsdl_idk)
mout.clear()
for k in [1, 2, 5, 10, 20]:
if k != 1:
val = knnestimate(data, data[fmeasure_idk]['input'], k)[klast:]
else:
val = knnestimate(data, data[fmeasure_idk]['input'], k)
klast = k
mout.append(["%d" % k, ", ".join(val)])
print("The %d nearest neighbors are:" % k, knnestimate(data, data[fmeasure_idk]['input'], k))
with open("tables/wsdl-knearest-stemmed.text", "w+") as out:
out.write(tabulate(mout, headers=headers, tablefmt="latex"))
def nonstemmeda8():
# load the data
blognames, words, data = readfile("datafiles/blogtop500.txt")
# find out where F-Measure is
fmeasure_idk = blognames.index('F-Measure')
# find out where wsdl is
wsdl_idk = blognames.index('Web Science and Digital Libraries Research Group')
# remove fmeasure from the data as we do not want to include it
takeout = data.pop(fmeasure_idk)
print("http://f-measure.blogspot.com/")
# hold the results of knn for reporting
mout = []
klast = None
# do knn
for k in [1, 2, 5, 10, 20]:
# for reporting I am shortening the output to only include the new data
# successive knn runs return the same data for previous k values
# ie k=2 [a,b] then k=5 would be [a,b,c,d,e] I only want k=2[a,b] and k=5[c,d,e]
if k != 1:
val = knnestimate(data, data[fmeasure_idk]['input'], k)[klast:]
else:
val = knnestimate(data, data[fmeasure_idk]['input'], k)
klast = k
mout.append(["%d" % k, ", ".join(val)])
print("The %d nearest neighbors are:" % k, knnestimate(data, data[fmeasure_idk]['input'], k))
# write out a table in latex
headers = ['k', 'nearest neighbors']
with open("tables/fmeasure-knearest-nonstemmed.text", "w+") as out:
out.write(tabulate(mout, headers=headers, tablefmt="latex"))
# do the same for wsdl
data.insert(fmeasure_idk, takeout)
print("http://ws-dl.blogspot.com/")
data.pop(wsdl_idk)
mout.clear()
for k in [1, 2, 5, 10, 20]:
if k != 1:
val = knnestimate(data, data[fmeasure_idk]['input'], k)[klast:]
else:
val = knnestimate(data, data[fmeasure_idk]['input'], k)
klast = k
mout.append(["%d" % k, ", ".join(val)])
print("The %d nearest neighbors are:" % k, knnestimate(data, data[fmeasure_idk]['input'], k))
headers = ['k', 'nearest neighbors']
with open("tables/wsdl-knearest-nonstemmed.text", "w+") as out:
out.write(tabulate(mout, headers=headers, tablefmt="latex"))
import numpredict
import clusters
if __name__ == "__main__":
blognames,words,data=clusters.readfile('blogdata.txt')
print 'F-Measure' # in data[75]
#print 'distance list: ' + str(numpredict.getdistances(data, blognames, data[75]))
print 'k = 1 ' + str(numpredict.knnestimate(data, blognames, data[75], k = 1))
print 'k = 2 ' + str(numpredict.knnestimate(data, blognames, data[75], k = 2))
print 'k = 5 ' + str(numpredict.knnestimate(data, blognames, data[75], k = 5))
print 'k = 10 ' + str(numpredict.knnestimate(data, blognames, data[75], k = 10))
print 'k = 20 ' + str(numpredict.knnestimate(data, blognames, data[75], k = 20))
print '\nWeb Science and Digital Libraries Research Group' # in data[7]
#print 'distance list: ' + str(numpredict.getdistances(data, blognames, data[7]))
print 'k = 1 ' + str(numpredict.knnestimate(data, blognames, data[7], k = 1))
print 'k = 2 ' + str(numpredict.knnestimate(data, blognames, data[7], k = 2))
print 'k = 5 ' + str(numpredict.knnestimate(data, blognames, data[7], k = 5))
print 'k = 10 ' + str(numpredict.knnestimate(data, blognames, data[7], k = 10))
print 'k = 20 ' + str(numpredict.knnestimate(data, blognames, data[7], k = 20))
def findNearestNeighbour(i, data, k):
testing = data[i]
neighbors = numpredict.knnestimate(data, testing, k)
for i in neighbors:
print(blogs[i[1]])
def knn1(d, v):
return numpredict.knnestimate(d, v, k=3)
import numpredict
# main function
# Building a Sample Dataset
print('\n<----Building a Sample Dataset--->')
print(numpredict.wineprice(90.0, 41.0))
data = numpredict.wineset1()
print(data[:2])
# K-Nearnest Neighbors
print('\n<----K-Nearnest Neighbors---->')
print(numpredict.knnestimate(data, (95.0, 3.0)))
print(numpredict.wineprice(95.0, 3.0))
print('<----Weighted KNN---->')
print(numpredict.weightedknn(data, (95.0, 3.0)))
# Cross_Validation
print('\n<----Cross-Validation---->')
data = numpredict.wineset1()
def knn1(d, v):
return numpredict.knnestimate(d, v, k=3)
def knn2(d, v):
return numpredict.weightedknn(d, v, k=3)
print(numpredict.crossvalidate(knn1, data))
print(numpredict.crossvalidate(knn2, data))
else:
return int(str_result)
def makeLaptopDataset():
searchResults = doSearch('laptop', categoryID=177)
result = []
for r in searchResults:
item = getItem(r['itemId'])
if item is None:
continue
att = item['attributes']
try:
data = (float(take_first_number(att['Memory'])), float(take_first_number(att['Hard Drive Capacity'])),
float(take_first_number(att['Screen Size'])),
float(item['feedback']))
print data
entry = {'input': data, 'result': float(item['price'])}
result.append(entry)
except:
print item['title']+' failed'
return result
data = makeLaptopDataset()
print 'Price=', numpredict.knnestimate(data, (8.0, 512.0, 15.0, 10000.0))
print 'Price=', numpredict.knnestimate(data, (8.0, 256.0, 15.0, 10000.0))
print 'Price=', numpredict.knnestimate(data, (8.0, 256.0, 13.0, 10000.0))
def knn1(d,v): return numpredict.knnestimate(d,v,k=1) def knninverse(d,v):return numpredict.weightedknn(d,v,weightf=numpredict.inverseweight)
def knn3(d,v): return numpredict.knnestimate(d,v,k=3) def knn1(d,v): return numpredict.knnestimate(d,v,k=1)
f = open("blogdata.txt", "r")
for line in f:
a = line.strip('\n').split('\t');
b = a.pop(0)
vecs[b] = a
fm = 'F-Measure'
ws = 'Web Science and Digital Libraries Research Group'
a = vecs[fm]
temp = vecs.values()
temp.pop(vecs.keys().index(fm))
a = numpredict.knnestimate(temp,a,k=5)
k = [1, 2, 5, 10, 20]
print "K Nearest Neighbors for F-Measure:"
for i in k:
print "When k = "+str(i)
for j in range(i):
b = a[j][1]
print vecs.keys()[b]
print '\n'
a = vecs[ws]
temp = vecs.values()
temp.pop(vecs.keys().index(ws))
attValue = getSingleValue(att, 'ValueLiteral')
attributes[attID] = attValue
result['attributes'] = attributes
return result
print getItem(laptops[7][0])
'''构造价格预测程序'''
def makeLaptopDataset():
searchResults = doSearch('laptop', categoryID=51148)
result = []
for r in searchResults:
item = getItem(r[0])
att = item['attributes']
try:
data = (float(att['12']), float(att['26444']), float(att['26446']),
float(att['25710']))
entry = {'input': data, 'result': float(item['price'])}
result.append(entry)
except:
print item['title'] + 'failed'
return result
set1 = makeLaptopDataset()
import numpredict
print numpredict.knnestimate(set1, (1024, 1000, 14, 40, 1000))
