def testMinChange(self):
optimizer = self._createPositiveOptimizer([5., 5.])
# This is more than enough to optimize the function
optimizer.maxLearningSteps = 10000
optimizer.minChange = 1e-6
result, f = optimizer.learn()
assertListAlmostEqual(self, result, [0., 0.], 0.001)
self.assertLess(optimizer.numLearningSteps, optimizer.maxLearningSteps)
def testDistribution(self):
input = [1, 2]
thetas = [0.5, 0.4, 0.3]
classifier = _LogisticRegression(thetas)
distribution = classifier.getDistribution(input)
posProbability = sigmoid(0.5 * 1 + 0.4 * 1 + 0.3 * 2)
expectedDistribution = [1 - posProbability, posProbability]
assertListAlmostEqual(self, distribution, expectedDistribution, 0.0001)
def testParaboloidOptimizationMaximize3(self):
optimizer = self._createNegativeOptimizer([-5., 5.])
result, f = optimizer.learn()
assertListAlmostEqual(self, result, [0., 0.], 0.001)
def testParaboloidOptimizationMinimize2(self):
optimizer = self._createPositiveOptimizer([5., -5.])
result, f = optimizer.learn()
assertListAlmostEqual(self, result, [0., 0.], 0.001)
