def test_resample_period(self):
"""
Test numerical and categorical labels with a resample period.
"""
cols = ['EEM', 'EWG', 'TIP', 'EWJ']
subset1 = self.data[cols].iloc[0:25]
subset2 = self.data[cols].iloc[0:100]
week_idx = subset1.resample('W').last().index
month_idx = subset2.resample('M').last().index
# Resample per week and per month.
test5 = excess_over_mean(subset1,
binary=False,
resample_by='W',
lag=True)
test6 = excess_over_mean(subset2,
binary=False,
resample_by='M',
lag=False)
test7 = excess_over_mean(subset2,
binary=True,
resample_by='M',
lag=False)
test5_actual = pd.DataFrame(
{
'EEM':
[0.017255, -0.042112, 0.004907, 0.016267, -0.026054, np.nan],
'EWG':
[-0.011975, -0.019257, -0.048906, 0.017310, -0.003467, np.nan],
'TIP':
[0.004898, 0.047419, 0.026425, -0.031773, 0.038212, np.nan],
'EWJ':
[-0.010178, 0.013950, 0.017574, -0.001804, -0.008691, np.nan]
},
index=week_idx)
test6_actual = pd.DataFrame(
{
'EEM': [np.nan, 0.018196, -0.027718, 0.047210, 0.001358],
'EWG': [np.nan, -0.008799, 0.026921, -0.007042, 0.007712],
'TIP': [np.nan, 0.0072872, 0.0037534, -0.069560, 0.0004006],
'EWJ': [np.nan, -0.016684, -0.002956, 0.029392, -0.009471]
},
index=month_idx)
pd.testing.assert_frame_equal(test5,
test5_actual,
check_less_precise=True)
pd.testing.assert_frame_equal(test6,
test6_actual,
check_less_precise=True)
pd.testing.assert_frame_equal(test7, test6_actual.apply(np.sign))
def test_large_set(self):
"""
Checks a specific row for a large dataset, and ensures the last row is NaN.
"""
# Get a random row of the entire dataset.
test3 = pd.Series(excess_over_mean(self.data, lag=True).iloc[42])
idx42 = self.data.iloc[42].index
test3_actual = pd.Series([
0.014267, 0.00537, -0.007375, -0.001013, 0.005282, -0.011075,
0.006921, 0.000626, 0.021376, 0.018429, -0.01463, -0.01377,
0.00048, -0.009569, 0.0107, 0.0121335, 0.0050173, 0.0002504,
-0.0004207, -0.018707, -0.013208, -0.00818807, -0.002897
],
index=idx42)
pd.testing.assert_series_equal(test3,
test3_actual,
check_less_precise=True,
check_names=False)
test4 = excess_over_mean(self.data, lag=True).iloc[-1]
self.assertTrue(test4.isnull().all())
def test_small_set(self):
"""
Check for a small set with manually inputted results, with numerical and categorical outputs.
"""
cols = ['EEM', 'EWG', 'TIP', 'EWJ']
subset = self.data[cols].iloc[0:5]
test1 = excess_over_mean(subset)
test2 = excess_over_mean(subset, binary=True)
test1_actual = pd.DataFrame([
(0.005666, -0.006157, 0.00004411, 0.0004476),
(-0.011169, -0.000684, 0.018588, -0.006734),
(0.006871, 0.000411, -0.000643, -0.006639),
(-0.003687, -0.002863, 0.004815, 0.001735),
(np.nan, np.nan, np.nan, np.nan),
],
columns=self.data[cols].iloc[0:5].columns,
index=self.data[cols].iloc[0:5].index)
test2_actual = test1_actual.apply(np.sign)
# Check less precise because calculated numbers have more decimal places than inputted ones.
pd.testing.assert_frame_equal(test1,
test1_actual,
check_less_precise=True)
pd.testing.assert_frame_equal(test2, test2_actual)