def test_prediction_values_middle_out_tdfp(self):
index = ['2000-01-01 00:00:00', '2000-01-01 00:00:01',
'2000-01-01 00:00:02', '2000-01-01 00:00:03',
'2000-01-01 00:00:04', '2000-01-01 00:00:05',
'2000-01-01 00:00:06', '2000-01-01 00:00:07',
'2000-01-01 00:00:08', '2000-01-01 00:00:09',
'2000-01-01 00:00:10', '2000-01-01 00:00:11',
'2000-01-01 00:00:12', '2000-01-01 00:00:13',
'2000-01-01 00:00:14', '2000-01-01 00:00:15',
'2000-01-01 00:00:16', '2000-01-01 00:00:17',
'2000-01-01 00:00:18', '2000-01-01 00:00:19']
time_series_model = AR(p=2, horizon=3)
data1_list = [1 for i in range(20)]
data2_list = [1 for i in range(10)]+[0 for i in range(10)]
data3_list = [0 for i in range(10)]+[1 for i in range(10)]
data1 = pd.DataFrame(index=index, data=data1_list)
data2 = pd.DataFrame(index=index, data=data2_list)
data3 = pd.DataFrame(index=index, data=data3_list)
data = {'data1': data1, 'data2': data2, 'data3': data3}
tree_adj = {'data1': ['data2', 'data3'], 'data2': [], 'data3': []}
values_prediction_data1 = [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]
values_prediction_child = [[0.0, 0.0, 0.0], [0.0, 0.0, 0.0], [0.0, 0.0, 0.0]]
test_prediction_data1 = pd.DataFrame(data=values_prediction_data1, columns=['y_1', 'y_2', 'y_3'], index=['2000-01-01 00:00:17', '2000-01-01 00:00:18', '2000-01-01 00:00:19'])
test_prediction_data2 = pd.DataFrame(data=values_prediction_child, columns=['y_1', 'y_2', 'y_3'], index=['2000-01-01 00:00:17', '2000-01-01 00:00:18', '2000-01-01 00:00:19'])
test_prediction_data3 = pd.DataFrame(data=values_prediction_data1, columns=['y_1', 'y_2', 'y_3'], index=['2000-01-01 00:00:17', '2000-01-01 00:00:18', '2000-01-01 00:00:19'])
prediction = HierarchicalMiddleOut(model=time_series_model, hierarchy_tree=tree_adj, root='data1', method='tdfp')
test_prediction_data = {'data1': test_prediction_data1, 'data2': test_prediction_data2, 'data3': test_prediction_data3}
prediction = prediction.fit(data).predict()
for key in test_prediction_data.keys():
assert_frame_equal(test_prediction_data[key], prediction[key])
def test_prediction_values_middle_out_middle_level(self):
index = ['2000-01-01 00:00:00', '2000-01-01 00:00:01',
'2000-01-01 00:00:02', '2000-01-01 00:00:03',
'2000-01-01 00:00:04', '2000-01-01 00:00:05',
'2000-01-01 00:00:06', '2000-01-01 00:00:07',
'2000-01-01 00:00:08', '2000-01-01 00:00:09',
'2000-01-01 00:00:10', '2000-01-01 00:00:11',
'2000-01-01 00:00:12', '2000-01-01 00:00:13',
'2000-01-01 00:00:14', '2000-01-01 00:00:15',
'2000-01-01 00:00:16', '2000-01-01 00:00:17',
'2000-01-01 00:00:18', '2000-01-01 00:00:19']
time_series_model = AR(p=2, horizon=3)
data1_list = [1 for i in range(20)]
data2_list = [1 for i in range(10)]+[0 for i in range(10)]
data3_list = [0 for i in range(10)]+[1 for i in range(10)]
data4_list = [0 for i in range(20)]
data5_list = [0.5 for i in range(20)]
data1 = pd.DataFrame(index=index, data=data1_list)
data2 = pd.DataFrame(index=index, data=data2_list)
data3 = pd.DataFrame(index=index, data=data3_list)
data4 = pd.DataFrame(index=index, data=data4_list)
data5 = pd.DataFrame(index=index, data=data5_list)
data = {'data1': data1, 'data2': data2, 'data3': data3, 'data4': data4, 'data5': data5}
tree_adj = {'data1': ['data2', 'data3'], 'data2': ['data4', 'data5'], 'data3': []}
prediction = HierarchicalMiddleOut(model=time_series_model, hierarchy_tree=tree_adj, root='data1', method='tdsga')
prediction.fit(data).predict(data)
def test_prediction_values_middle_out_middle_level(self):
index = [
"2000-01-01 00:00:00",
"2000-01-01 00:00:01",
"2000-01-01 00:00:02",
"2000-01-01 00:00:03",
"2000-01-01 00:00:04",
"2000-01-01 00:00:05",
"2000-01-01 00:00:06",
"2000-01-01 00:00:07",
"2000-01-01 00:00:08",
"2000-01-01 00:00:09",
"2000-01-01 00:00:10",
"2000-01-01 00:00:11",
"2000-01-01 00:00:12",
"2000-01-01 00:00:13",
"2000-01-01 00:00:14",
"2000-01-01 00:00:15",
"2000-01-01 00:00:16",
"2000-01-01 00:00:17",
"2000-01-01 00:00:18",
"2000-01-01 00:00:19",
]
time_series_model = AR(p=2, horizon=3)
data1_list = [1 for i in range(20)]
data2_list = [1 for i in range(10)] + [0 for i in range(10)]
data3_list = [0 for i in range(10)] + [1 for i in range(10)]
data4_list = [0 for i in range(20)]
data5_list = [0.5 for i in range(20)]
data1 = pd.DataFrame(index=index, data=data1_list)
data2 = pd.DataFrame(index=index, data=data2_list)
data3 = pd.DataFrame(index=index, data=data3_list)
data4 = pd.DataFrame(index=index, data=data4_list)
data5 = pd.DataFrame(index=index, data=data5_list)
data = {
"data1": data1,
"data2": data2,
"data3": data3,
"data4": data4,
"data5": data5,
}
tree_adj = {
"data1": ["data2", "data3"],
"data2": ["data4", "data5"],
"data3": [],
}
prediction = HierarchicalMiddleOut(
model=time_series_model,
hierarchy_tree=tree_adj,
root="data1",
method="tdsga",
)
prediction.fit(data).predict(data)
def test_results(self, time_series, p, horizon):
ar = AR(p, horizon)
predictions = ar.fit(time_series).predict()
assert predictions.shape[0] == horizon
assert predictions.shape[1] == horizon
def test_constructor(self, p, horizon):
ar = AR(p, horizon)
assert len(ar.features) == p
assert ar.horizon == horizon
def test_prediction_values_middle_out_tdfp(self):
index = [
"2000-01-01 00:00:00",
"2000-01-01 00:00:01",
"2000-01-01 00:00:02",
"2000-01-01 00:00:03",
"2000-01-01 00:00:04",
"2000-01-01 00:00:05",
"2000-01-01 00:00:06",
"2000-01-01 00:00:07",
"2000-01-01 00:00:08",
"2000-01-01 00:00:09",
"2000-01-01 00:00:10",
"2000-01-01 00:00:11",
"2000-01-01 00:00:12",
"2000-01-01 00:00:13",
"2000-01-01 00:00:14",
"2000-01-01 00:00:15",
"2000-01-01 00:00:16",
"2000-01-01 00:00:17",
"2000-01-01 00:00:18",
"2000-01-01 00:00:19",
]
time_series_model = AR(p=2, horizon=3)
data1_list = [1 for i in range(20)]
data2_list = [1 for i in range(10)] + [0 for i in range(10)]
data3_list = [0 for i in range(10)] + [1 for i in range(10)]
data1 = pd.DataFrame(index=index, data=data1_list)
data2 = pd.DataFrame(index=index, data=data2_list)
data3 = pd.DataFrame(index=index, data=data3_list)
data = {"data1": data1, "data2": data2, "data3": data3}
tree_adj = {"data1": ["data2", "data3"], "data2": [], "data3": []}
values_prediction_data1 = [[1.0, 1.0, 1.0], [1.0, 1.0, 1.0], [1.0, 1.0, 1.0]]
values_prediction_child = [[0.0, 0.0, 0.0], [0.0, 0.0, 0.0], [0.0, 0.0, 0.0]]
test_prediction_data1 = pd.DataFrame(
data=values_prediction_data1,
columns=["y_1", "y_2", "y_3"],
index=["2000-01-01 00:00:17", "2000-01-01 00:00:18", "2000-01-01 00:00:19"],
)
test_prediction_data2 = pd.DataFrame(
data=values_prediction_child,
columns=["y_1", "y_2", "y_3"],
index=["2000-01-01 00:00:17", "2000-01-01 00:00:18", "2000-01-01 00:00:19"],
)
test_prediction_data3 = pd.DataFrame(
data=values_prediction_data1,
columns=["y_1", "y_2", "y_3"],
index=["2000-01-01 00:00:17", "2000-01-01 00:00:18", "2000-01-01 00:00:19"],
)
prediction = HierarchicalMiddleOut(
model=time_series_model,
hierarchy_tree=tree_adj,
root="data1",
method="tdfp",
)
test_prediction_data = {
"data1": test_prediction_data1,
"data2": test_prediction_data2,
"data3": test_prediction_data3,
}
prediction = prediction.fit(data).predict()
for key in test_prediction_data.keys():
assert_frame_equal(test_prediction_data[key], prediction[key])
def test_features_are_correct(self, time_series, p, horizon):
ar = AR(p, horizon)
for i, feature in enumerate(ar.features):
assert feature[0] == f"s{i}"
assert isinstance(feature[1], Shift)
assert feature[1].shift == i
def test_constructor(self, p, horizon, explainer_type):
ar = AR(p, horizon, explainer_type=explainer_type)
assert len(ar.features) == p
assert ar.horizon == horizon
assert ar.explainer_type == explainer_type
assert ar.model.explainer_type == explainer_type
