def testMF(W, H, trainingDict, testDict, n):
t = test.MFtest(W, H, evalDict)
test.hitrate(testDict, t.getRec, n)
def constant(r, trainingDict, testDict):
rec = baselines.constant(trainingDict)
print("Hitrate for constant: %r" %
test.hitrate(testDict, rec.getRec, 10))
def random(r, trainingDict, testDict):
rec = baselines.randomRec(trainingDict)
print("Hitrate for random: %r" % test.hitrate(testDict, rec.getRec, 10))
# #reg = 0.01 # Regularization constant #ler = 0.1 # Learning rate #features = 150 # number of features #EPOCHS = 3 # number of epochs # #W, H = mf.learnModel( # r.getMaxUid(), r.getMaxIid(), # reg, reg, reg, ler, r.getR(), # features, EPOCHS, # r.numberOfTransactions, # mf.logLoss, mf.dLogLoss) # With logLoss it will be BPRMF # ## Get recommendations with a MF model. #import test #import external #t = test.MFtest(W, H, r.getR()) #getRec = external.getExternalRec(t.getRec, r) #print getRec(97, 10) # Calculate the hitrate of item based cosine KNN. import split splittedMatrix, testDict = split.splitMatrix(r.getMatrix(), 1234567890) import knn neighbors = 10 # number of neighbors k = knn.itemKnn(splittedMatrix, neighbors) import test print test.hitrate(testDict, k.getRec, 10)
