def myclassify_oneclass(numfiers,xtrain,xtest,xtrunc,nuparam=.5):
# remove NaN, Inf, and -Inf values from the xtest feature matrix
xtest = xtest[~np.isnan(xtest).any(axis=1),:]
xtest = xtest[~np.isinf(xtest).any(axis=1),:]
xtrunc = xtrunc[0] # this is because the xtrunc is a list and we want only the first index, which is our array
#if xtest is NxM matrix, returns Nxnumifiers matrix where each column corresponds to a classifiers prediction vector
count = 0
# print numfiers
predictionMat = np.empty((xtest.shape[0],numfiers))
predictionStringMat = []
finalPredMat = []
oneclassifier = oneclass(nu = nuparam)
# print 'created classifier'
oneclassifier = oneclassifier.fit(xtrain)
# print 'trained classifier'
ytest = oneclassifier.predict(xtest)
# print 'predicted classifier results'
predictionMat[:,count] = ytest
count += 1
for colCount in range(predictionMat.shape[1]):
tempCol = predictionMat[:,colCount]
modeCol = predWindowVecModeFinder(tempCol,xtrunc)
finalPredMat += map(int,modeCol)
modeStr = predVec2Str(modeCol)
predictionStringMat.append(modeStr)
return predictionStringMat,finalPredMat
xtrain_pow_shortened = xtrain_pow[:,0:6]
tuned_params = [{'kernel': ['rbf'],'tol' : [.9,1]}] #,{'kernel': ['linear']},{'kernel': ['poly'], 'degree': [2,6],'gamma':[.1,.0001],'coef0': [0.0,1.0]}]
#indep_grid ={'iid': [True,False]} then use ParameterGrid?
oc = oneclass()
oc.fit(xtrain_aud_shortened)
predVec = oc.predict(xtrain_aud_shortened)
predVec = (predVec+1)/2
for i in range(len(predVec)):
predVec[i] = int(predVec[i])
score = recsco(np.ones(xtrain_aud_shortened.shape),predVec)
print "Score %s" % score
#gridSearch = GridSearchCV(estimator = oneclass(),param_grid = tuned_params,iid=True,scoring=recsco)
print "starting grid search"
#gridSearch.fit(xtrain_aud)
print("Best params found")
# print()
def myclassify_NA(numfiers,xtrain,xtest,xtrunc,xtest2,xtrunc2,nuparam):
# remove NaN, Inf, and -Inf values from the xtest feature matrix
xtest = xtest[~np.isnan(xtest).any(axis=1),:]
xtest = xtest[~np.isinf(xtest).any(axis=1),:]
xtrunclength = sio.loadmat('Files/xtrunclength.mat')
xtrunclength = xtrunclength['xtrunclength'][0]
xtrunc = xtrunc[0]
xtrunc2 = xtrunc2[0]
#if xtest is NxM matrix, returns Nxnumifiers matrix where each column corresponds to a classifiers prediction vector
count = 0
print numfiers
predictionMat = np.empty((xtest.shape[0],numfiers))
predictionStringMat = []
predictionMat2 = np.empty((xtest2.shape[0],numfiers))
predictionStringMat2 = []
print 'part 1'
oneclassclass = oneclass(nu = nuparam)
print 'part 2'
oneclassclass.fit(xtrain)
print 'part 3'
ytest = oneclassclass.predict(xtest)
ytest2 = oneclassclass.predict(xtest2)
print 'part 4'
predictionMat[:,count] = ytest
predictionMat2[:,count] = ytest2
count += 1
#print oneclassclass.get_params()
# print 'finished one'
# if count < numfiers:
# oc2 = oneclass(nu = .05)
# oc2.fit(xtrain)
# ytest = oc2.predict(xtest)
# predictionMat[:,count] = ytest
# count +=1
# #
# if count < numfiers:
# oc3 = oneclass(kernel='linear', degree = 4)
# oc3.fit(xtrain)
# ytest = oc3.predict(xtest)
# predictionMat[:,count] = ytest
# count+=1
#
# if count < numfiers:
# oneclass2 = oneclass(kernel = 'poly', degree = 4)
# print 'part 5'
# oneclass2.fit(xtrain)
# print 'part 6'
# ytest = oneclass2.predict(xtest)
# print 'part 7'
# predictionMat[:,count] = ytest
# count+=1
if count < numfiers:
oneclass2 = oneclass(kernel = 'poly', nu = nuparam)
print 'part 5'
oneclass2.fit(xtrain)
print 'part 6'
ytest = oneclass2.predict(xtest)
ytest2 = oneclass2.predict(xtest2)
print 'part 7'
predictionMat[:,count] = ytest
predictionMat2[:,count] = ytest2
count+=1
for colCount in range(predictionMat.shape[1]):
tempCol = predictionMat[:,colCount]
modeCol = predWindowVecModeFinder(tempCol,xtrunc)
modeStr = predVec2Str(modeCol)
predictionStringMat.append(modeStr)
for colCount2 in range(predictionMat2.shape[1]):
tempCol2 = predictionMat2[:,colCount2]
modeCol2 = predWindowVecModeFinder(tempCol2,xtrunc2)
modeStr2 = predVec2Str(modeCol2)
predictionStringMat2.append(modeStr2)
return predictionStringMat,predictionStringMat2
