def WMulti(files,
file,
testSet,
fout,
features,
name,
clfName,
dp,
convertToBinary=True):
train = []
for file2 in files:
if (file2[0:3] == file[0:3] and file2 < file < 0):
train.append(file2)
if (len(train)):
trainSet = DPLIB.LoadCSV(train, dp, features, convertToBinary)
if (name.lower().find("infogain") >= 0):
#int indi[] = DPLIB.fSelectInfoGain(trainSet);
#if (DPLIB.useIterativeInfoGainSubsetting)
#{
# indi = DPLIB.iterativeInfoGainSubsetting(trainSet, indi,clfName);
#}
#else
# indi = DPLIB.getTopX(indi);
#trainSet = DPLIB.fSelectSet(trainSet, indi);
#testSet = DPLIB.fSelectSet(testSet, indi);
pass
l = GLOB(clfName).getClassifier()
l.buildClassifier(trainSet)
vec = l.evaluateModel(testSet)
tvals = DPLIB.getConfMatrix(testSet[:, -1], vec)
auc = DPLIB.getAUC(testSet[:, -1], vec)
vals = DPLIB.getMeasures(tvals)
print(name + ":" + file + ": " + str(vals) + " AUC = " + str(auc))
fout.write("\n" + name + ":" + file + ": " + " AUC = " + str(auc) +
";" + "Vals=" + str(vals))
else:
print(name + ":" + file + ": " + "!!!" + " AUC = !!!")
fout.write("\n" + name + ":" + file + ": !!!")
def run(self):
lrnrnames = self.lrnrs
try:
rnd = random.Random(Common.getCurrentTimeMil())
if self.expType == 'GIS':
if self.isKS:
self.gis = GISKS2(self.pars,self.file)
else:
self.gis = GIS(self.pars)
elif self.expType == 'LSH':
lsh = CPDP_LSH_Binary(self.pars, self.file)
trainSetAll = DPLIB.LoadCSV(self.train, self.dp, self.features, convertToBinary = not self.isCount);
testSetAll = DPLIB.LoadCSV(self.test, self.dp, self.features, convertToBinary = not self.isCount);
ft = 'A'
indi = None
if not self.isCount:
if self.pars['features'] == 'Iterative InfoGain Subsetting':
ft = 'IG'
indi = DPLIB.fSelectInfoGain(trainSetAll);
if self.pars['features'] == 'All':
print ('All')
if self.pars['features'] == 'PCA':
ft = 'PCA'
print ('PCA')
trainSetAll, testSetAll = DPLIB.applyPCA(trainSetAll, testSetAll, 0.95)
for lk in range(len(lrnrnames)):
lrnr = "-" + lrnrnames[lk];
clfName = lrnrnames[lk];
vSets = None
if not self.isCount:
if self.pars['features'] == 'Iterative InfoGain Subsetting':
indis2 = DPLIB.iterativeInfoGainSubsetting(trainSetAll, indi, clfName);
trainSetAll = DPLIB.fSelectSet(trainSetAll, indis2);
testSetAll = DPLIB.fSelectSet(testSetAll, indis2);
if self.pars['vSetType'] in ['Single Random','Multiple Random']:
vSets = DatasetUtils.getRandomDatasets()
c = 0
while (c < self.iters):
print("Start:" + self.file + ": " + str(c));
print("====================================================");
#fout.write("#ITERINFO:For File=" +self.file + "-Iter:" + str(c) + "\n");
stages = None
buckets = None
sbtx = ""
if self.expType == 'GIS':
self.doGIS(trainSetAll, testSetAll, "FIXED-VMUL-GEN-"+ft, lrnr, fout, vSets, False, clfName, gis=gis);
gis.prnt('---------------------------------------\n')
elif self.expType == 'LSH':
lsh.CreateBucketsTune(trainSetAll, testSetAll, vSets, name= "LSHTune-ALL-TOP-SUPER" + sbtx + lrnr, testCut= self.pars['tunecut'], iternum=c, save=False, superbit=self.pars['lshType'] =='SuperBit', clfName=clfName,tunelrn = self.pars['tunelrnr']);
lsh.prnt('---------------------------------------\n')
c+=1
#fout.write("===================================================================\n");
#fout.close();
print("File Processing Ended:" +self.file);
except Exception as e:
try:
print (str(e))
print(traceback.format_exc())
except Exception as ex2:
print("X2", str(ex2));
print(traceback.format_exc())
if self.expType == 'GIS':
return gis
elif self.expType == 'LSH':
return lsh