def frequencyOnLabels(trainFile, modify, flag, userNum=339, wsNum=5825, sparess=2):
import numpy as np
from paper.rendi2 import loadDataset as ld
trainData = ld.loadTestList(trainFile, modify=modify)
if trainData.shape[1] == 4:
trainData = trainData[:, 1:]
userFreq = np.zeros((userNum, 21))
wsFreq = np.zeros((wsNum, 21))
for index, line in enumerate(trainData):
if flag == True and index % sparess != lineNum:
continue
user, ws, tui = line
user = int(user)
ws = int(ws)
labelIndex = int(tui) + 1 #labelIndex i means label i-1
userFreq[user, labelIndex] += 1
wsFreq[ws, labelIndex] += 1
return userFreq, wsFreq, trainData
def doMapping(trainFile, testFile, modify=True, flag=True, sparess=10):
global userIps
# import numpy as np
from paper.rendi2 import loadDataset as ld
userFreq, wsFreq, trainData = frequencyOnLabels(trainFile,
modify=modify,
flag=flag,
sparess=sparess)
#trainFeature
result = []
for index, line in enumerate(trainData):
temp = []
user, ws, tui = line
user = int(user)
ws = int(ws)
temp.extend(userFreq[user])
temp.extend(userIps[user])
temp.extend(wsFreq[ws])
temp.append(tui)
result.append(temp)
trainFeature = np.array(result, dtype=float)
#testFeature
result = []
testData = ld.loadTestList(testFile, modify=modify)
for index, line in enumerate(testData):
temp = []
user, ws, tui = line
user = int(user)
ws = int(ws)
temp.extend(userFreq[user])
temp.extend(userIps[user])
temp.extend(wsFreq[ws])
temp.append(tui)
result.append(temp)
testFeature = np.array(result, dtype=float)
return trainFeature, testFeature
import time
from paper.rendi2 import cfPredictByDBSCAN
from paper.rendi2 import cfPredictByEuclid
start = time.time()
for sparess in [5, 10, 15, 20]:
for fileNum in range(1, 11):
print sparess, fileNum, '\t',
#load the qos dataset
trainFile = "dataset/rendi2/train/sparseness%s/training%d.txt" % (
sparess, fileNum)
simFile1 = "dataset/rendi2/train/sparseness%s/training%deuSimMatrix" % (
sparess, fileNum)
simFile2 = "dataset/rendi2/train/sparseness%s/training%dcooSimMatrix" % (
sparess, fileNum)
trainArray = ld.loadArrayObj(trainFile)
trainData = ld.loadTestList(trainFile)
simArrayEu = np.loadtxt(simFile1)
simArrayCoo = np.loadtxt(simFile2)
#load the prob dataset
kProb = 1
trainProb = "dataset/rendi2/train/sparseness%s/training%d-prob" % (
sparess, fileNum)
trainProbData = np.loadtxt(trainProb)
labels = np.argsort(trainProbData,
axis=1)[:, -kProb:] - 1 #top-k labels
labelsProb = np.sort(trainProbData,
axis=1)[:, -kProb:] #top-k labels's
#cal eui
euiProb = 0.0
count = 0.0
pui = []
if __name__ == "__main__":
from paper.rendi2 import loadDataset as ld
import numpy as np
import time
start = time.time()
for sparess in [5, 10, 15, 20]:
for fileNum in range(1, 11):
#load the qos dataset
trainFile = "dataset/rendi2/train/sparseness%s/training%d.txt" % (
sparess, fileNum)
testFile = "dataset/rendi2/test/sparseness%s/test%d.txt" % (
sparess, fileNum)
simFile = "dataset/rendi2/train/sparseness%s/training%deuSimMatrix" % (
sparess, fileNum)
trainArray = ld.loadArrayObj(trainFile)
testData = ld.loadTestList(testFile)
simArray = np.loadtxt(simFile)
#load the prob dataset
kProb = 1
testProb = "dataset/rendi2/test/sparseness%s/test%d-prob" % (
sparess, fileNum)
testProbData = np.loadtxt(testProb)
labels = np.argsort(testProbData,
axis=1)[:, -kProb:] - 1 #top-k labels
labelsProb = np.sort(testProbData,
axis=1)[:, -kProb:] #top-k labels's
#cal eui
euiProb = 0.0
count = 0.0
pui = []
k = 2
@author: root
"""
from paper.rendi2 import loadDataset as ld
from sklearn.ensemble import AdaBoostClassifier, AdaBoostRegressor, RandomForestRegressor, RandomForestClassifier
from sklearn.model_selection import GridSearchCV
from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor
from sklearn.naive_bayes import GaussianNB
import numpy as np
import time
start = time.time()
#load the prob dataset
trainProb = "dataset/rendi2/training-0501-after-prob"
testProb = "dataset/rendi2/test-0501-after-prob"
x = ld.loadTestList(trainProb)
testX = ld.loadTestList(testProb)
#load the qos dataset
train = "dataset/rendi2/training1.txt"
test = "dataset/rendi2/test1.txt"
y = ld.loadTestList(train)[:, -1]
testY = ld.loadTestList(test)[:, -1]
clf = RandomForestRegressor(oob_score=True,
n_jobs=20,
n_estimators=100,
max_features=0.1,
min_samples_split=10)
clf.fit(x, y.astype(float))
return None
return (np.mat(rating) * np.mat(numbers).T)[0, 0] / sums
if __name__ == '__main__':
import time
from paper.rendi2 import loadDataset as ld
import numpy as np
start = time.time()
NoneValue = 111111
userNum = 339
wsNum = 5825
for sparess in [5]:
for fileNum in range(1, 2):
# sampleTrainFile = "dataset/rendi2/sample/training-%d-%d" % (sparess, fileNum)
trainFile = "dataset/rendi2/train/sparseness%s/training%d.txt" % (
sparess, fileNum)
testFile = "dataset/rendi2/puiAnalyze/puiAnalyze-%d-%d" % (sparess,
fileNum)
simFile = "dataset/rendi2/train/sparseness%s/training%deuSimMatrix" % (
sparess, fileNum)
simArray = np.loadtxt(simFile)
trainArray = ld.loadArrayObj(trainFile, modify=True)
testData = ld.loadTestList(testFile, modify=False)
calMaeAndRmse(trainArray, testData, simArray)
#end modeling
print "during time ... ", time.time() - start