def test_get_rolling_timeseries__window_size_one__same_dataframe_is_returned(
self):
test_df = self.get_test_dataframe()
expected_df = test_df
rolled_df = get_rolling_timeseries(df_x=test_df,
start_index=0,
lag=0,
window_size=1)
print(rolled_df)
assert len(rolled_df) == len(expected_df)
assert_frame_equal(expected_df, rolled_df[rolled_df.columns[:-1]])
def test_get_rolling_timeseries__increasing_window_size__rolling_is_done_correct(
self):
test_df = self.get_test_dataframe()
for i in range(2, len(test_df)):
rolled_df = get_rolling_timeseries(df_x=test_df,
start_index=0,
lag=0,
window_size=i)
assert (len(test_df) - i + 1) * i == len(rolled_df)
# check if the window ids are set correctly
assert (test_df.index[:-i +
1] == rolled_df.window_id.unique()).all()
# check if all subgrousps have the correct size
assert (rolled_df.groupby('window_id').size() == i).all()
def test_get_rolling_timeseries__lag_one_window_size_one__same_dataframe_except_last_row_is_returned_and_window_id_starts_at_1(
self):
test_df = self.get_test_dataframe()
expected_df = test_df[:-1]
rolled_df = get_rolling_timeseries(df_x=test_df,
start_index=0,
lag=1,
window_size=1)
print(rolled_df)
assert len(rolled_df) == len(expected_df)
assert_frame_equal(expected_df, rolled_df[rolled_df.columns[:-1]])
assert (list(range(1,
len(expected_df) +
1)) == rolled_df['window_id']).all()
def test_get_rolling_timeseries__changing_start_index_over_zero__sub_dataframe_is_returned(
self):
test_df = self.get_test_dataframe()
for i in range(1, len(test_df)):
rolled_df = get_rolling_timeseries(df_x=test_df,
start_index=i,
lag=0,
window_size=1)
print(rolled_df)
print(test_df.index[:-i])
assert len(test_df) - i == len(rolled_df)
assert (test_df.index[i:] == rolled_df.window_id).all()
assert (test_df.A[i:].values == rolled_df.A.values).all()
assert (test_df.B[i:].values == rolled_df.B.values).all()
assert (test_df.C[i:].values == rolled_df.C.values).all()