def test_mae_mse(self):
data = [{'id': 'AA', 'hr': 1, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 2, 'temp': 0.5, 'dist': 0, 'altitude': 0.5},
{'id': 'CC', 'hr': 3, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data)
self.assertAlmostEqual(inst_regression.get_mae(), 0)
self.assertAlmostEqual(inst_regression.get_mse(), 0)
def test_threshold(self):
data = [{'id': 'AA', 'hr': 0, 'temp': 0.3, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 0.4, 'dist': 0.1,
'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 1, 'dist': 0.4, 'altitude': 1}
]
inst_regression = MultiRegression(data, score_threshold=0.000000001)
self.assertEqual(len(set(inst_regression.used_vars)
.intersection(['dist', 'altitude'])), 2)
inst_regression = MultiRegression(data, score_threshold=0.5)
self.assertEqual(len(set(inst_regression.used_vars)
.intersection(['dist', 'altitude'])), 1)
def test_predict_points(self):
data = [{'id': 'AA', 'hr': 1, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 2, 'temp': 0.5, 'dist': 0, 'altitude': 0.5},
{'id': 'CC', 'hr': 3, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data)
results = inst_regression.predict_points(
[{'dist': 11, 'altitude': 22}, {'dist': 11, 'altitude': 23}])
self.assertAlmostEqual(results[0], 22)
self.assertAlmostEqual(results[1], 23)
def test_regression_ideal_data(self):
data = [{'id': 'AA', 'hr': 0, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 0.5, 'dist': 0,
'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegressionSigma(data, sigma_limit=1.5)
# print(vars(inst_regression))
coefs = inst_regression.get_coefs()
self.assertEqual(len(coefs), 2)
self.assertEqual(len(coefs[0]), 1)
self.assertAlmostEqual(coefs[0][0], 1)
self.assertAlmostEqual(coefs[1], 0)
self.assertAlmostEqual(inst_regression.get_score(), 1)
data = []
i = 0
for altitude in range(0, 2):
for hr in range(0, 2):
for dist in range(0, 2):
i += 1
data.append({'id': 'key_' + str(i),
'hr': hr, 'dist': dist,
'altitude': altitude,
'temp': 1 + hr + 2 * dist + 3 * altitude
})
inst_regression = MultiRegression(data,
x_vars=['altitude', 'dist', 'hr'])
# print(vars(inst_regression))
coefs1 = inst_regression.get_coefs()
data.append({'id': 'BAD', 'hr': 0, 'temp': 0, 'dist': 0,
'altitude': 0})
inst_regression = MultiRegressionSigma(data, sigma_limit=1.5,
x_vars=['altitude',
'dist', 'hr'])
# print(vars(inst_regression))
coefs2 = inst_regression.get_coefs()
# The MultiRegressionSigma class must have filtered the bad data
self.assertEqual(len(coefs2[0]), len(coefs1[0]))
self.assertAlmostEqual(coefs2[0][0], coefs1[0][0])
self.assertAlmostEqual(coefs2[1], coefs1[1])
# The MultiRegressionSigma must have the result as if the bad element
# didn't exist.
self.assertEqual(len(coefs2), 2)
self.assertEqual(len(coefs2[0]), 3)
self.assertAlmostEqual(coefs2[0][0], 3)
self.assertAlmostEqual(coefs2[1], 1)
self.assertAlmostEqual(inst_regression.get_score(), 1)
# The Multiregression sigma must return the residuals for all the data,
# not only the one inside the sigma values
self.assertAlmostEqual(len(data), len(inst_regression.get_residuals()))
inst_regression.predict_points([{'id': 'AA', 'hr': 0, 'temp': 0,
'dist': 0, 'altitude': 0}])
inst_regression.regr.predict([[0, 0, 0]])
def test_predict_point(self):
data = [{'id': 'AA', 'hr': 1, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 2, 'temp': 0.5, 'dist': 0, 'altitude': 0.5},
{'id': 'CC', 'hr': 3, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data)
self.assertAlmostEqual(inst_regression.predict_point(
{'dist': 11, 'altitude': 22}), 22)
# Three variable correlation & score 1 y = 1 + a + 2b + 3c
data = [{'id': 'AA', 'hr': 0, 'temp': 9, 'dist': 4, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 9, 'dist': 3, 'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 9, 'dist': 2, 'altitude': 1},
{'id': 'DD', 'hr': 1, 'temp': 8, 'dist': 1, 'altitude': 2},
{'id': 'EE', 'hr': 4, 'temp': 17, 'dist': 0.5, 'altitude': 3},
{'id': 'FF', 'hr': 2, 'temp': 11, 'dist': 0, 'altitude': 4},
{'id': 'GG', 'hr': 2, 'temp': 12, 'dist': 0, 'altitude': 5},
{'id': 'HH', 'hr': 2, 'temp': 14, 'dist': 1, 'altitude': 5}
]
inst_regression = MultiRegression(data, y_var='temp',
x_vars=['altitude', 'dist', 'hr'])
self.assertAlmostEqual(
inst_regression.predict_point(
{'dist': 11, 'altitude': 22, 'hr': 1}), 48)
def test_get_residuals(self):
data = [{'id': 'AA', 'hr': 1, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 2, 'temp': 0.5, 'dist': 0, 'altitude': 0.5},
{'id': 'CC', 'hr': 3, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data)
residuals = inst_regression.get_residuals()
self.assertTrue(isinstance(residuals, dict))
for key in residuals.keys():
self.assertAlmostEqual(residuals[key], 0)
data = [{'id': 'AA', 'hr': 0, 'temp': 9, 'dist': 4, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 7.5, 'dist': 3,
'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 6, 'dist': 2, 'altitude': 1},
{'id': 'DD', 'hr': 1, 'temp': 5, 'dist': 1, 'altitude': 2},
{'id': 'EE', 'hr': 1, 'temp': 5, 'dist': 0.5, 'altitude': 3},
{'id': 'FF', 'hr': 1, 'temp': 5, 'dist': 0, 'altitude': 4}
]
inst_regression = MultiRegression(data)
residuals = inst_regression.get_residuals()
self.assertTrue(isinstance(residuals, dict))
for key in residuals.keys():
self.assertAlmostEqual(residuals[key], 0)
# Make the correlation not be perfect so some residuals exist
data = [{'id': 'AA', 'hr': 0, 'temp': 0, 'dist': 4, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 9, 'dist': 3, 'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 9, 'dist': 2, 'altitude': 1},
{'id': 'DD', 'hr': 1, 'temp': 8, 'dist': 1, 'altitude': 2},
{'id': 'EE', 'hr': 4, 'temp': 17, 'dist': 0.5, 'altitude': 3},
{'id': 'FF', 'hr': 2, 'temp': 11, 'dist': 0, 'altitude': 4},
{'id': 'GG', 'hr': 2, 'temp': 12, 'dist': 0, 'altitude': 5},
{'id': 'HH', 'hr': 2, 'temp': 14, 'dist': 1, 'altitude': 5}
]
inst_regression = MultiRegression(data)
residuals = inst_regression.get_residuals()
self.assertTrue(isinstance(residuals, dict))
for key in residuals.keys():
self.assertTrue(abs(residuals[key]) > 0)
def test_change_input_values(self):
data = [{'name': 'AA', 'hr': 0, 'temp': 0, 'dist': 0, 'altitude': 0},
{'name': 'BB', 'hr': 0.5, 'temp': 0.5, 'dist': 0,
'altitude': 0.5},
{'name': 'CC', 'hr': 1, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data, id_key='name')
residuals = inst_regression.get_residuals()
self.assertTrue('BB' in residuals)
data = [{'id': 'AA', 'hr': 0, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 0.5, 'dist': 0,
'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data, y_var='hr')
coefs = inst_regression.get_coefs()
self.assertEqual(len(coefs), 2)
self.assertEqual(len(coefs[0]), 1)
self.assertAlmostEqual(coefs[0][0], 1)
data = [{'id': 'AA', 'hr': 0, 'temp': 0, 'dist': 0, 'hhh': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 0.5, 'dist': 0,
'hhh': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 1, 'dist': 0, 'hhh': 1}
]
inst_regression = MultiRegression(data, x_vars=['dist', 'hhh'])
coefs = inst_regression.get_coefs()
self.assertEqual(len(coefs), 2)
self.assertEqual(len(coefs[0]), 1)
self.assertAlmostEqual(coefs[0][0], 1)
def test_regression_ideal_data(self):
# Correlation with altitude and coef = 1
data = [{'id': 'AA', 'hr': 0, 'temp': 0, 'dist': 0, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 0.5, 'dist': 0,
'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 1, 'dist': 0, 'altitude': 1}
]
inst_regression = MultiRegression(data)
coefs = inst_regression.get_coefs()
self.assertEqual(len(coefs), 2)
self.assertEqual(len(coefs[0]), 1)
self.assertAlmostEqual(coefs[0][0], 1)
self.assertAlmostEqual(coefs[1], 0)
self.assertAlmostEqual(inst_regression.get_score(), 1)
# Correlation with two variables & score 1 y = 1 + a + 2b
data = [{'id': 'AA', 'hr': 0, 'temp': 9, 'dist': 4, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 7.5, 'dist': 3,
'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 6, 'dist': 2, 'altitude': 1},
{'id': 'DD', 'hr': 1, 'temp': 5, 'dist': 1, 'altitude': 2},
{'id': 'EE', 'hr': 1, 'temp': 5, 'dist': 0.5, 'altitude': 3},
{'id': 'FF', 'hr': 1, 'temp': 5, 'dist': 0, 'altitude': 4}
]
inst_regression = MultiRegression(data)
coefs = inst_regression.get_coefs()
self.assertEqual(len(coefs), 2)
self.assertEqual(len(coefs[0]), 2)
self.assertAlmostEqual(coefs[0][0], 2)
self.assertAlmostEqual(coefs[0][1], 1)
self.assertAlmostEqual(coefs[1], 1)
self.assertAlmostEqual(inst_regression.get_score(), 1)
# Correlation with three variables & score 1 y = 1 + a + 2b + 3c
data = [{'id': 'AA', 'hr': 0, 'temp': 9, 'dist': 4, 'altitude': 0},
{'id': 'BB', 'hr': 0.5, 'temp': 9, 'dist': 3, 'altitude': 0.5},
{'id': 'CC', 'hr': 1, 'temp': 9, 'dist': 2, 'altitude': 1},
{'id': 'DD', 'hr': 1, 'temp': 8, 'dist': 1, 'altitude': 2},
{'id': 'EE', 'hr': 4, 'temp': 17, 'dist': 0.5, 'altitude': 3},
{'id': 'FF', 'hr': 2, 'temp': 11, 'dist': 0, 'altitude': 4},
{'id': 'GG', 'hr': 2, 'temp': 12, 'dist': 0, 'altitude': 5},
{'id': 'HH', 'hr': 2, 'temp': 14, 'dist': 1, 'altitude': 5}
]
inst_regression = MultiRegression(data, y_var='temp',
x_vars=['altitude', 'dist', 'hr'])
coefs = inst_regression.get_coefs()
self.assertEqual(len(coefs), 2)
self.assertEqual(len(coefs[0]), 3)
self.assertAlmostEqual(coefs[0][0], 3)
self.assertAlmostEqual(coefs[0][1], 2)
self.assertAlmostEqual(coefs[0][2], 1)
self.assertAlmostEqual(coefs[1], 1)
self.assertAlmostEqual(inst_regression.get_score(), 1)
# Let's worsen the correlation
data[0]['temp'] = 0
inst_regression = MultiRegression(data, y_var='temp',
x_vars=['altitude', 'dist', 'hr'])
self.assertTrue(inst_regression.get_score() < 1)