def is_poisson(data_array):
max_window = np.max(data_array, axis=0)
frac = max_window / data_array.sum()
pval = chisq_poisson(np.array(data_array, 'i'))[1]
if pval <= SIGNIFICANCE: #Not poisson
return (False, frac)
else:
return (True, frac)
def test_example_three_chisq(self):
'''Same as one but with non contiguous data'''
data = [8] * 2 + [7] * 5 + [6] * 10 + [5] * 21 + \
[4] * 29 + [3] * 41 + [2] * 47 + [1] * 31 + [0] * 14
random.shuffle(data)
result = chisq_poisson(data)
self.assertAlmostEqual(2.28, result[0], 2)
self.assertAlmostEqual(0.94, result[1], 2)
def test_example_two_chisq(self):
'''
Based on the example:
http://mlsc.lboro.ac.uk/resources/statistics/gofit.pdf
(we do not treat minimum frequencies as the authors does)
'''
data = [3] * 4 + [2] * 9 + [1] * 15 + [0] * 32
result = chisq_poisson(data)
self.assertAlmostEqual(3.46, result[0], 2)
self.assertAlmostEqual(0.18, result[1], 2)
def test_example_one_chisq(self):
'''
Based on the example:
http://goo.gl/PO0U8
(we do not treat minimum frequencies as the authors does)
'''
data = [8] * 2 + [7] * 5 + [6] * 10 + [5] * 21 + \
[4] * 29 + [3] * 41 + [2] * 47 + [1] * 31 + [0] * 14
result = chisq_poisson(data)
self.assertAlmostEqual(2.28, result[0], 2)
self.assertAlmostEqual(0.94, result[1], 2)
