def testAucFromRoc(self):
self.treeRank = TreeRankR()
self.treeRank.setLeafRank(self.treeRank.getTreeRankLib().LRsvm)
self.treeRank.learnModel(self.X, self.Y)
roc = self.treeRank.predictROC(self.X, self.Y)
auc = self.treeRank.aucFromROC(roc)
self.assertAlmostEquals(auc, 0.7626141, 2)
self.assertTrue(0 <= auc <= 1)
def setUp(self):
logging.basicConfig(stream=sys.stdout, level=logging.DEBUG)
self.baseLib = importr('base')
self.treeRankLib = importr('TreeRank')
utilLib = importr("utils")
utilLib.data("TRdata")
XY = numpy.array(robjects.r["Gauss2D.learn"]).T
self.X = XY[:, 0:2]
self.Y = numpy.array([XY[:, 2]], numpy.int).T
trainExamples = 500
self.trainX = self.X[0:trainExamples, :]
self.trainY = self.Y[0:trainExamples, :]
self.testX = self.X[trainExamples:, :]
self.testY = self.Y[trainExamples:, :]
self.treeRank = TreeRankR()
def profileLearnModel(self):
treeRank = TreeRankR()
treeRank.printMemStats = True
treeRank.setMaxDepth(2)
treeRank.setMinSplit(50)
treeRank.setLeafRank(treeRank.getLrLinearSvmPlain())
numExamples = 650
numFeatures = 950
X = numpy.random.rand(numExamples, numFeatures)
Y = numpy.array(numpy.random.rand(numExamples) < 0.1, numpy.int)
def run():
for i in range(5):
print("Iteration " + str(i))
treeRank.learnModel(X, Y)
#print(treeRank.getTreeSize())
#print(treeRank.getTreeDepth())
ProfileUtils.profile('run()', globals(), locals())
def profileLearnModel(self):
treeRank = TreeRankR()
treeRank.printMemStats = True
treeRank.setMaxDepth(2)
treeRank.setMinSplit(50)
treeRank.setLeafRank(treeRank.getLrLinearSvmPlain())
numExamples = 650
numFeatures = 950
X = numpy.random.rand(numExamples, numFeatures)
Y = numpy.array(numpy.random.rand(numExamples) < 0.1, numpy.int)
def run():
for i in range(5):
print("Iteration " + str(i))
treeRank.learnModel(X, Y)
#print(treeRank.getTreeSize())
#print(treeRank.getTreeDepth())
ProfileUtils.profile('run()', globals(), locals())
def testInit(self):
self.treeRank = TreeRankR()
class TreeRankRTest(unittest.TestCase):
def setUp(self):
logging.basicConfig(stream=sys.stdout, level=logging.DEBUG)
self.baseLib = importr('base')
self.treeRankLib = importr('TreeRank')
utilLib = importr("utils")
utilLib.data("TRdata")
XY = numpy.array(robjects.r["Gauss2D.learn"]).T
self.X = XY[:, 0:2]
self.Y = numpy.array([XY[:, 2]], numpy.int).T
trainExamples = 500
self.trainX = self.X[0:trainExamples, :]
self.trainY = self.Y[0:trainExamples, :]
self.testX = self.X[trainExamples:, :]
self.testY = self.Y[trainExamples:, :]
self.treeRank = TreeRankR()
def testInit(self):
self.treeRank = TreeRankR()
#def testLearnModelDataFrame(self):
#self.treeRank.learnModelDataFrame(self.formula, self.XY)
def testPredictROC(self):
self.treeRank.learnModel(self.trainX, self.trainY)
roc = self.treeRank.predictROC(self.trainX, self.trainY)
self.assertEquals(roc.shape[1], 2)
#Compare predictROC to predict + the scikits roc function
scores = self.treeRank.predict(self.trainX)
fpr, tpr, threshold = learn.metrics.roc_curve(self.trainY.ravel(), scores)
roc2 = numpy.c_[fpr, tpr]
roc2 = numpy.r_[numpy.zeros((1, 2)), roc2]
tol = 10**-6
self.assertTrue(numpy.linalg.norm(roc - roc2) < tol)
#Now try on test set
roc = self.treeRank.predictROC(self.testX, self.testY)
scores = self.treeRank.predict(self.testX)
fpr, tpr, threshold = learn.metrics.roc_curve(self.testY.ravel(), scores)
roc2 = numpy.c_[fpr, tpr]
roc2 = numpy.r_[numpy.zeros((1, 2)), roc2]
self.assertTrue(numpy.linalg.norm(roc - roc2) < tol)
def testPredictScores(self):
self.treeRank.learnModel(self.X, self.Y)
scores = self.treeRank.predictScores(self.X)
self.assertEquals(scores.shape[0], self.X.shape[0])
def testAucFromRoc(self):
self.treeRank = TreeRankR()
self.treeRank.setLeafRank(self.treeRank.getTreeRankLib().LRsvm)
self.treeRank.learnModel(self.X, self.Y)
roc = self.treeRank.predictROC(self.X, self.Y)
auc = self.treeRank.aucFromROC(roc)
self.assertAlmostEquals(auc, 0.7626141, 2)
self.assertTrue(0 <= auc <= 1)
def testLearnModel(self):
numExamples = 10
numFeatures = 5
X = numpy.random.rand(numExamples, numFeatures)
Y = numpy.random.rand(numExamples)
self.treeRank.learnModel(X, Y)
@unittest.skip("Demonstrating skipping")
def testEvaluateCv(self):
folds = 3
mean, var = self.treeRank.evaluateCv(self.X, self.Y.ravel(), folds, metricMethod=TreeRankR.auc)
#Compare versus just train/test splits
trainExamples = 500
trainX = self.X[0:trainExamples, :]
trainY = self.Y[0:trainExamples, :]
testX = self.X[trainExamples:, :]
testY = self.Y[trainExamples:, :]
self.treeRank.learnModel(trainX, trainY)
scores = self.treeRank.predict(testX)
#print(TreeRankR.auc(scores, testY))
self.assertTrue(0 <= mean <= 1)
self.assertTrue(0 <= var <= 1)
def testAuc(self):
self.treeRank.learnModel(self.X, self.Y)
scores = self.treeRank.predictScores(self.X)
auc1 = Evaluator.auc(scores, self.Y.ravel())
auc2 = self.treeRank.aucFromROC(self.treeRank.predictROC(self.X, self.Y))
self.assertAlmostEquals(auc1, auc2, places=4)
#@unittest.skip("Demonstrating skipping")
def testEvaluateCvOuter(self):
folds = 3
leafRank = self.treeRank.getTreeRankLib().LRsvm
trainExamples = 200
trainX = self.X[0:trainExamples, :]
trainY = self.Y[0:trainExamples, :]
bestParams, allMetrics, bestMetaDicts = self.treeRank.evaluateCvOuter(trainX, trainY.ravel(), folds, leafRank)
for i in range(folds):
logging.debug(allMetrics[2][i])
logging.debug(bestParams[i])
def testGetTreeSize(self):
self.treeRank.learnModel(self.X, self.Y)
self.assertTrue(self.treeRank.getTreeSize() > 0)
def testGetTreeDepth(self):
self.treeRank.learnModel(self.X, self.Y)
self.assertTrue(self.treeRank.getTreeDepth() > 0)
def testGetLRsvm2(self):
LRsvm2 = self.treeRank.getLrRbfSvm()
self.treeRank.setLeafRank(LRsvm2)
self.treeRank.learnModel(self.X, self.Y)
Y = self.treeRank.predict(self.X)
def testGetLRCart2(self):
LRCart2 = self.treeRank.getLrCart()
self.treeRank.setLeafRank(LRCart2)
self.treeRank.learnModel(self.X, self.Y)
Y = self.treeRank.predict(self.X)
def learnModel(self, X, Y):
Parameter.checkClass(X, numpy.ndarray)
Parameter.checkClass(Y, numpy.ndarray)
Parameter.checkArray(X)
Parameter.checkArray(Y)
if numpy.unique(Y).shape[0] < 2:
raise ValueError("Vector of labels must be binary, currently numpy.unique(Y) = " + str(numpy.unique(Y)))
#If Y is 1D make it 2D
if Y.ndim == 1:
Y = numpy.array([Y]).T
XY = self._getDataFrame(X, Y)
formula = robjects.Formula('class ~ .')
self.learnModelDataFrame(formula, XY)
if self.printMemStats:
logging.debug(self.getLsos()())
logging.debug(ProfileUtils.memDisplay(locals()))
