def test_compute_traffic_light_bounds():
datastore = {
"test_data": pytest.test_comparer_df,
}
conf = {
"monitoring_rules": {
"the_feature:mae": [8, 4, 2, 2],
"dummy_feature:*": [0, 0, 0, 0],
"mse": [0.2, 0.11, 0.09, 0],
"mae": [0, 0, 0, 0],
"*": [0, 0, 0, 0],
}
}
module = ComputeTLBounds(
read_key="test_data",
store_key="output_data",
ignore_features=["dummy_feature"],
monitoring_rules=conf["monitoring_rules"],
)
output = module.transform(datastore)["output_data"]
assert "dummy_feature:mae" not in output.keys()
assert output["the_feature:mae"] == [8, 4, 2, 2]
assert output["the_feature:mse"] == [0.2, 0.11, 0.09, 0]
def test_compute_traffic_light_funcs():
datastore = {"test_data": pytest.test_comparer_df}
conf = {
"monitoring_rules": {
"the_feature:mae": [8, 4, 2, 2],
"dummy_feature:*": [0, 0, 0, 0],
"mse": [0.2, 0.11, 0.09, 0],
"mae": [0, 0, 0, 0],
"*": [0, 0, 0, 0],
}
}
module = ComputeTLBounds(
read_key="test_data",
apply_funcs_key="output_data",
monitoring_rules=conf["monitoring_rules"],
)
output = module.transform(datastore)["output_data"]
assert len(output) == 3
assert output[0]["features"] == ["dummy_feature"]
assert output[0]["metrics"] == ["mae"]
assert output[0]["args"] == (0, 0, 0, 0)
assert output[1]["features"] == ["the_feature"]
assert output[1]["metrics"] == ["mae"]
assert output[1]["args"] == (8, 4, 2, 2)
assert output[2]["features"] == ["the_feature"]
assert output[2]["metrics"] == ["mse"]
assert output[2]["args"] == (0.2, 0.11, 0.09, 0)
def test_report_traffic_light_bounds():
datastore = dict()
datastore["to_profile"] = {"asc_numbers": get_test_data()}
conf = {
"monitoring_rules": {
"the_feature:mae": [8, 4, 2, 0.15],
"mse": [0.2, 0.11, 0.09, 0],
"mae": [1, 0, 0, 0],
"*_pull": [7, 4, -4, -7]
},
"pull_rules": {
"*_pull": [7, 4, -4, -7]
}
}
m1 = ApplyFunc(apply_to_key="to_profile",
features=["asc_numbers"],
metrics=['a', 'b'])
m1.add_apply_func(expanding_mean, suffix='_std', entire=True)
m1.add_apply_func(expanding_std, suffix='_mean', entire=True)
m2 = ApplyFunc(apply_to_key="to_profile", features=["asc_numbers"])
m2.add_apply_func(pull,
suffix='_pull',
axis=1,
suffix_mean='_mean',
suffix_std='_std')
ctlb = ComputeTLBounds(
read_key="to_profile",
store_key="static_tlb",
monitoring_rules=conf["monitoring_rules"],
)
m3 = ComputeTLBounds(read_key="to_profile",
monitoring_rules=conf["pull_rules"],
apply_funcs_key="dynamic_tlb",
func=pull_bounds,
metrics_wide=True,
axis=1)
m4 = ApplyFunc(
apply_to_key=m3.read_key,
assign_to_key='dtlb',
apply_funcs_key="dynamic_tlb",
)
rg = SectionGenerator(read_key="to_profile",
store_key="section",
section_name="Profiles",
dynamic_bounds='dtlb',
static_bounds='static_tlb')
pipeline = Pipeline(modules=[m1, m2, ctlb, m3, m4, rg])
datastore = pipeline.transform(datastore)
def test_traffic_light_summary_combination():
datastore = {"test_data": test_comparer_df}
conf = {
"monitoring_rules": {
"the_feature:mae": [8, 4, 2, 0.15],
"dummy_feature:*": [0, 0, 0, 0],
"mse": [0.2, 0.11, 0.09, 0],
"mae": [0, 0, 0, 0],
"*": [0, 0, 0, 0],
}
}
ctlb = ComputeTLBounds(
read_key="test_data",
store_key="traffic_light_bounds",
apply_funcs_key="traffic_light_funcs",
ignore_features=["dummy_feature"],
monitoring_rules=conf["monitoring_rules"],
prefix="tl_",
)
atlb = ApplyFunc(
apply_to_key=ctlb.read_key,
assign_to_key="output_data",
apply_funcs_key="traffic_light_funcs",
)
tls = ApplyFunc(
apply_to_key="output_data",
apply_funcs=[dict(func=traffic_light_summary, axis=1, suffix="")],
assign_to_key="alerts",
)
asum = AlertsSummary(read_key="alerts")
pipeline = Pipeline(modules=[ctlb, atlb, tls, asum])
datastore = pipeline.transform(datastore)
alerts = datastore["alerts"]
assert "_AGGREGATE_" in alerts
output = datastore["alerts"]["_AGGREGATE_"]
assert output["worst"].values[-1] == 2
assert output["n_green"].values[-1] == 1
assert output["n_yellow"].values[-1] == 0
assert output["n_red"].values[-1] == 1
def test_integration_alerting():
datastore = {
"test_data": test_comparer_df,
}
conf = {
"monitoring_rules": {
"the_feature:mae": [8, 4, 2, 0.15],
"dummy_feature:*": [0, 0, 0, 0],
"mse": [0.2, 0.11, 0.09, 0],
"mae": [0, 0, 0, 0],
"*": [0, 0, 0, 0],
}
}
ctlb = ComputeTLBounds(
read_key="test_data",
store_key="traffic_light_bounds",
apply_funcs_key="traffic_light_funcs",
ignore_features=["dummy_feature"],
monitoring_rules=conf["monitoring_rules"]
)
atlb = ApplyFunc(
apply_to_key=ctlb.read_key,
assign_to_key='output_data',
apply_funcs_key="traffic_light_funcs",
)
pipeline = Pipeline(modules=[ctlb, atlb])
datastore = pipeline.transform(datastore)
output = datastore[atlb.store_key]["the_feature"]
alerts_per_color_per_date = pd.DataFrame()
for i, color in enumerate(["green", "yellow", "red"]):
alerts_per_color_per_date[f"n_{color}"] = (output.values == i).sum(axis=1)
alerts_total_per_color = alerts_per_color_per_date.sum(axis=0)
assert alerts_total_per_color["n_green"] == 5
assert alerts_total_per_color["n_yellow"] == 1
assert alerts_total_per_color["n_red"] == 4
def test_traffic_light_summary():
datastore = {
"test_data": test_comparer_df,
}
conf = {
"monitoring_rules": {
"the_feature:mae": [8, 4, 2, 0.15],
"dummy_feature:*": [0, 0, 0, 0],
"mse": [0.2, 0.11, 0.09, 0],
"mae": [0, 0, 0, 0],
"*": [0, 0, 0, 0],
}
}
ctlb = ComputeTLBounds(
read_key="test_data",
store_key="traffic_light_bounds",
apply_funcs_key="traffic_light_funcs",
ignore_features=["dummy_feature"],
monitoring_rules=conf["monitoring_rules"],
prefix='tl_'
)
atlb = ApplyFunc(
apply_to_key=ctlb.read_key,
assign_to_key='output_data',
apply_funcs_key="traffic_light_funcs",
)
tls = ApplyFunc(apply_to_key='output_data', apply_funcs=[dict(func=traffic_light_summary, axis=1, suffix='')],
assign_to_key='alerts')
pipeline = Pipeline(modules=[ctlb, atlb, tls])
datastore = pipeline.transform(datastore)
output = datastore['alerts']["the_feature"]
assert output["worst"].values[-1] == 2
assert output["n_green"].values[-1] == 1
assert output["n_yellow"].values[-1] == 0
assert output["n_red"].values[-1] == 1