def testMH(self):
data = pd.DataFrame({
"Clicks": [1, 3, 2, 3, 1, 2],
"Conversions": [1, 0, 1, 2, 1, 1],
"Id": [1, 2, 3, 1, 2, 3],
"Condition": [0, 0, 0, 1, 1, 1]
})
weights = np.ones(6)
comparison = comparisons.MH("Condition", 0, "Id")
comparison.precalculate_factors(data)
metric = metrics.Ratio("Conversions", "Clicks")
output = comparison(data, weights, metric).values[0]
ka = np.array([2, 1, 1])
kb = np.array([1, 0, 1])
na = np.array([3, 1, 2])
nb = np.array([1, 3, 2])
w = 1 / (na + nb)
correct = sum(ka * nb * w) / sum(kb * na * w)
self.assertEqual(output, correct)
def testRatioWithWeights(self):
df = pd.DataFrame({"X": [1, 1, 1, 2, 3, 4], "Y": [1, 2, 0, 1, 1, 1]})
weights = np.array([1, 1, 1, 2, 1, 1])
metric = metrics.Ratio("X", "Y")
output = metric(df, weights)
correct = 2.0
self.assertEqual(output, correct)
def testRatioZeroDivideByZero(self):
df = pd.DataFrame({"X": [0, 0], "Y": [0, 0]})
weights = np.array([1, 1])
metric = metrics.Ratio("X", "Y")
output = metric(df, weights)
self.assertTrue(np.isnan(output))
def testRatioPositiveDivideByZero(self):
df = pd.DataFrame({"X": [1, 1], "Y": [0, 0]})
weights = np.array([1, 1])
metric = metrics.Ratio("X", "Y")
output = metric(df, weights)
correct = np.inf
self.assertEqual(output, correct)
def testJackknifeRatio(self):
data = pd.DataFrame({"X": [1, 2, 3, 4], "Y": [4, 3, 2, 1]})
metric = metrics.Ratio("X", "Y")
se_method = standard_errors.Jackknife()
output = core.Analyze(data).with_standard_errors(se_method).calculate(
metric).run()
estimates = np.array([9 / 6, 8 / 7, 7 / 8, 6 / 9])
rss = ((estimates - estimates.mean())**2).sum()
se = np.sqrt(rss * 3 / 4)
correct = pd.DataFrame([[1.0, se]],
columns=("X/Y", "X/Y Jackknife SE"))
self.assertTrue(output.equals(correct))
def testMHZeroRateRaisesError(self):
data = pd.DataFrame({
"Clicks": [2, 1, 1, 0, 2, 0, 0, 0, 0, 0, 0, 0],
"Conversions": [1, 1, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0],
"Id": [1, 1, 2, 2, 3, 3, 1, 1, 2, 2, 3, 3],
"Condition": [0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1]
})
weights = np.ones(12)
comparison = comparisons.MH("Condition", 0, "Id")
comparison.precalculate_factors(data)
metric = metrics.Ratio("Conversions", "Clicks")
with self.assertRaises(ValueError):
comparison(data, weights, metric)