def test_cum(self):
# create new column (cumsum)
post_df = proc.cum(df=timeseries_df, columns={"y": "y2"}, operator="sum",)
self.assertListEqual(post_df.columns.tolist(), ["label", "y", "y2"])
self.assertListEqual(series_to_list(post_df["label"]), ["x", "y", "z", "q"])
self.assertListEqual(series_to_list(post_df["y"]), [1.0, 2.0, 3.0, 4.0])
self.assertListEqual(series_to_list(post_df["y2"]), [1.0, 3.0, 6.0, 10.0])
# overwrite column (cumprod)
post_df = proc.cum(df=timeseries_df, columns={"y": "y"}, operator="prod",)
self.assertListEqual(post_df.columns.tolist(), ["label", "y"])
self.assertListEqual(series_to_list(post_df["y"]), [1.0, 2.0, 6.0, 24.0])
# overwrite column (cummin)
post_df = proc.cum(df=timeseries_df, columns={"y": "y"}, operator="min",)
self.assertListEqual(post_df.columns.tolist(), ["label", "y"])
self.assertListEqual(series_to_list(post_df["y"]), [1.0, 1.0, 1.0, 1.0])
# invalid operator
self.assertRaises(
QueryObjectValidationError,
proc.cum,
df=timeseries_df,
columns={"y": "y"},
operator="abc",
)
def test_cum_should_not_side_effect():
_timeseries_df = timeseries_df.copy()
pp.cum(
df=timeseries_df,
columns={"y": "y2"},
operator="sum",
)
assert _timeseries_df.equals(timeseries_df)
def test_cum_with_pivot_df_and_multiple_metrics(self):
pivot_df = proc.pivot(
df=multiple_metrics_df,
index=["dttm"],
columns=["country"],
aggregates={
"sum_metric": {
"operator": "sum"
},
"count_metric": {
"operator": "sum"
},
},
flatten_columns=False,
reset_index=False,
)
cum_df = proc.cum(
df=pivot_df,
operator="sum",
is_pivot_df=True,
)
# dttm count_metric, UK count_metric, US sum_metric, UK sum_metric, US
# 0 2019-01-01 1 2 5 6
# 1 2019-01-02 4 6 12 14
assert cum_df["count_metric, UK"].to_list() == [1.0, 4.0]
assert cum_df["count_metric, US"].to_list() == [2.0, 6.0]
assert cum_df["sum_metric, UK"].to_list() == [5.0, 12.0]
assert cum_df["sum_metric, US"].to_list() == [6.0, 14.0]
assert (cum_df["dttm"].to_list() == to_datetime([
"2019-01-01",
"2019-01-02",
]).to_list())
def test_cum_with_pivot_df_and_single_metric(self):
pivot_df = proc.pivot(
df=single_metric_df,
index=["dttm"],
columns=["country"],
aggregates={"sum_metric": {
"operator": "sum"
}},
flatten_columns=False,
reset_index=False,
)
cum_df = proc.cum(
df=pivot_df,
operator="sum",
is_pivot_df=True,
)
# dttm UK US
# 0 2019-01-01 5 6
# 1 2019-01-02 12 14
assert cum_df["UK"].to_list() == [5.0, 12.0]
assert cum_df["US"].to_list() == [6.0, 14.0]
assert (cum_df["dttm"].to_list() == to_datetime([
"2019-01-01",
"2019-01-02",
]).to_list())
def test_cum_after_pivot_with_multiple_metrics():
pivot_df = pp.pivot(
df=multiple_metrics_df,
index=["dttm"],
columns=["country"],
aggregates={
"sum_metric": {
"operator": "sum"
},
"count_metric": {
"operator": "sum"
},
},
flatten_columns=False,
reset_index=False,
)
"""
count_metric sum_metric
country UK US UK US
dttm
2019-01-01 1 2 5 6
2019-01-02 3 4 7 8
"""
cum_df = pp.cum(
df=pivot_df,
operator="sum",
columns={
"sum_metric": "sum_metric",
"count_metric": "count_metric"
},
)
"""
count_metric sum_metric
country UK US UK US
dttm
2019-01-01 1 2 5 6
2019-01-02 4 6 12 14
"""
flat_df = pp.flatten(cum_df)
"""
dttm count_metric, UK count_metric, US sum_metric, UK sum_metric, US
0 2019-01-01 1 2 5 6
1 2019-01-02 4 6 12 14
"""
assert flat_df.equals(
pd.DataFrame({
"dttm": pd.to_datetime(["2019-01-01", "2019-01-02"]),
FLAT_COLUMN_SEPARATOR.join(["count_metric", "UK"]): [1, 4],
FLAT_COLUMN_SEPARATOR.join(["count_metric", "US"]): [2, 6],
FLAT_COLUMN_SEPARATOR.join(["sum_metric", "UK"]): [5, 12],
FLAT_COLUMN_SEPARATOR.join(["sum_metric", "US"]): [6, 14],
}))
def test_cum():
# create new column (cumsum)
post_df = pp.cum(
df=timeseries_df,
columns={"y": "y2"},
operator="sum",
)
assert post_df.columns.tolist() == ["label", "y", "y2"]
assert series_to_list(post_df["label"]) == ["x", "y", "z", "q"]
assert series_to_list(post_df["y"]) == [1.0, 2.0, 3.0, 4.0]
assert series_to_list(post_df["y2"]) == [1.0, 3.0, 6.0, 10.0]
# overwrite column (cumprod)
post_df = pp.cum(
df=timeseries_df,
columns={"y": "y"},
operator="prod",
)
assert post_df.columns.tolist() == ["label", "y"]
assert series_to_list(post_df["y"]) == [1.0, 2.0, 6.0, 24.0]
# overwrite column (cummin)
post_df = pp.cum(
df=timeseries_df,
columns={"y": "y"},
operator="min",
)
assert post_df.columns.tolist() == ["label", "y"]
assert series_to_list(post_df["y"]) == [1.0, 1.0, 1.0, 1.0]
# invalid operator
with pytest.raises(InvalidPostProcessingError):
pp.cum(
df=timeseries_df,
columns={"y": "y"},
operator="abc",
)
def test_cum_after_pivot_with_single_metric():
pivot_df = pp.pivot(
df=single_metric_df,
index=["dttm"],
columns=["country"],
aggregates={"sum_metric": {
"operator": "sum"
}},
flatten_columns=False,
reset_index=False,
)
"""
sum_metric
country UK US
dttm
2019-01-01 5 6
2019-01-02 7 8
"""
cum_df = pp.cum(df=pivot_df,
operator="sum",
columns={"sum_metric": "sum_metric"})
"""
sum_metric
country UK US
dttm
2019-01-01 5 6
2019-01-02 12 14
"""
cum_and_flat_df = pp.flatten(cum_df)
"""
dttm sum_metric, UK sum_metric, US
0 2019-01-01 5 6
1 2019-01-02 12 14
"""
assert cum_and_flat_df.equals(
pd.DataFrame({
"dttm": pd.to_datetime(["2019-01-01", "2019-01-02"]),
FLAT_COLUMN_SEPARATOR.join(["sum_metric", "UK"]): [5, 12],
FLAT_COLUMN_SEPARATOR.join(["sum_metric", "US"]): [6, 14],
}))
