numpy.random.seed(21)
logging.debug(maxLocalAuc)
#maxLocalAuc.learningRateSelect(trainX)
U, V, trainMeasures, testMeasures, iterations, time = maxLocalAuc.learnModel(trainX, U=U, V=V, verbose=True)
fprTrain, tprTrain = MCEvaluator.averageRocCurve(trainX, U, V)
fprTest, tprTest = MCEvaluator.averageRocCurve(testX, U, V)
return fprTrain, tprTrain, fprTest, tprTest
if saveResults:
paramList = []
chunkSize = 1
U, V = maxLocalAuc.initUV(X)
for loss, rho in losses:
for trainX, testX in trainTestXs:
maxLocalAuc.loss = loss
maxLocalAuc.rho = rho
paramList.append((trainX, testX, maxLocalAuc.copy(), U.copy(), V.copy()))
pool = multiprocessing.Pool(maxtasksperchild=100, processes=multiprocessing.cpu_count())
resultsIterator = pool.imap(computeTestAuc, paramList, chunkSize)
#import itertools
#resultsIterator = itertools.imap(computeTestAuc, paramList)
meanFprTrains = []
meanTprTrains = []
def computeObjectives(args):
trainX, maxLocalAuc, U, V = args
numpy.random.seed(21)
logging.debug(maxLocalAuc)
U, V, trainMeasures, testMeasures, iterations, time = maxLocalAuc.learnModel(trainX, U=U, V=V, verbose=True)
return trainMeasures[-1, 0]
if saveResults:
#First run with low learning rate to get a near-optimal solution
U, V = maxLocalAuc.initUV(trainX)
maxLocalAuc.maxIterations = 5000
U2, V2, trainMeasures, testMeasures, iterations, time = maxLocalAuc.learnModel(trainX, U=U, V=V, verbose=True)
idealTrainMeasures = trainMeasures[:, 0]
maxLocalAuc.maxIterations = 100
paramList = []
objectives1 = numpy.zeros((t0s.shape[0], alphas.shape[0], etas.shape[0], startAverages.shape[0], folds))
for t0 in t0s:
for alpha in alphas:
for eta in etas:
for startAverage in startAverages:
for trainX, testX in trainTestXs: