def multiclass_on_binary_svms(Xnew, Y, testXnew, testY): print "Classifying with a one vs one SVM" classify(Xnew, Y, testXnew, testY, True, 0.2) print "Classifying with a one vs all SVM" classify(Xnew, Y, testXnew, testY, False, 0.2)
def monb_crossvalidate(X, Y, c, onevone): iters = select_cross_validation_subsets(X, Y, 6) #crange = [0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, # 1.0, 10.0, 100.0, 1000.0] crange = [1.0] for c in crange: print "C=%f"%c for d in iters: X = d["train"][0] Y = d["train"][1] testX = d["test"][0] testY = d["test"][1] classify(X, Y, testX, testY, onevone, c)
def monb_test(X, Y, c, onevone, ntest): X, Y, testX, testY = select_test_set(X, Y, ntest) Y = map(lambda v: v*1.0, Y) classify(X, Y, testX, testY, onevone, c)