maxLocalAuc.numProcesses = multiprocessing.cpu_count()
maxLocalAuc.numRecordAucSamples = 100
maxLocalAuc.numRowSamples = 30
maxLocalAuc.rate = "constant"
maxLocalAuc.recordStep = 10
maxLocalAuc.rho = 1.0
maxLocalAuc.t0 = 1.0
maxLocalAuc.t0s = 2.0**-numpy.arange(7, 12, 1)
maxLocalAuc.validationSize = 3
maxLocalAuc.validationUsers = 0
newM = X.shape[0]/4
modelSelectX, userInds = Sampling.sampleUsers(X, newM)
if saveResults:
meanObjs1, stdObjs1 = maxLocalAuc.modelSelect(X)
meanObjs2, stdObjs2 = maxLocalAuc.modelSelect(trainX)
meanObjs3, stdObjs3 = maxLocalAuc.modelSelect(modelSelectX)
numpy.savez(outputFile, meanObjs1, meanObjs2, meanObjs3)
else:
data = numpy.load(outputFile)
meanObjs1, meanObjs2, meanObjs3 = data["arr_0"], data["arr_1"], data["arr_2"]
meanObjs1 = numpy.squeeze(meanObjs1)
meanObjs2 = numpy.squeeze(meanObjs2)
meanObjs3 = numpy.squeeze(meanObjs3)
import matplotlib
matplotlib.use("GTK3Agg")
X5, userInds = Sampling.sampleUsers2(X, 500000, prune=False)
X6, userInds = Sampling.sampleUsers2(X, 200000, prune=False)
X7, userInds = Sampling.sampleUsers2(X, 100000, prune=False)
print(X.shape, X.nnz)
print(X2.shape, X2.nnz)
print(X3.shape, X3.nnz)
print(X4.shape, X4.nnz)
print(X5.shape, X5.nnz)
print(X6.shape, X6.nnz)
print(X7.shape, X7.nnz)
meanF1s1, stdF1s1 = maxLocalAuc.modelSelect(X)
meanF1s2, stdF1s2 = maxLocalAuc.modelSelect(X2)
meanF1s3, stdF1s3 = maxLocalAuc.modelSelect(X3)
meanF1s4, stdF1s4 = maxLocalAuc.modelSelect(X4)
meanF1s5, stdF1s5 = maxLocalAuc.modelSelect(X5)
meanF1s6, stdF1s6 = maxLocalAuc.modelSelect(X6)
meanF1s7, stdF1s7 = maxLocalAuc.modelSelect(X7)
numpy.savez(outputFile, meanF1s1, meanF1s2, meanF1s3, meanF1s4, meanF1s5, meanF1s6, meanF1s7)
else:
data = numpy.load(outputFile)
meanF1s1, meanF1s2, meanF1s3, meanF1s4, meanF1s5, meanF1s6, meanF1s7 = data["arr_0"], data["arr_1"], data["arr_2"], data["arr_3"], data["arr_4"], data["arr_5"], data["arr_6"]
meanF1s1 = numpy.squeeze(meanF1s1)
