def get_code_id_for_package_name(self, df, code_system_id):
assert "target_package" in df, \
"Input data must have a 'target_package' column."
cause_map = (engine_room.get_cause_map(
code_system_id,
**self.standard_cache_options).loc[:, ["value", "code_id"]])
cause_map["value"] = misc.clean_icd_codes(cause_map.value, True)
pkg = (
engine_room.get_package_list(code_system_id,
include_garbage_codes=True,
**self.standard_cache_options).
pipe(engine_room.remove_five_plus_digit_icd_codes,
"garbage_code",
code_system_id=code_system_id).assign(
value=lambda d: misc.clean_icd_codes(
d.garbage_code, True)).loc[:,
["package_name", "value"]]
# pick first garbage code found in the package
.drop_duplicates("package_name", keep="first").merge(
cause_map, how="left").loc[:, ["package_name", "code_id"]])
df = df.merge(pkg,
left_on="target_package",
right_on="package_name",
how="left")
# assert that we found a code for each package in target_package
# where target_package is not NaN
misc.report_if_merge_fail(df[df.target_package.notna()], "code_id",
"target_package")
return df
def make_cause_fractions(self, df):
if 'mean_env' not in df.columns:
df = add_envelope(df, env_df=self.env_df)
report_if_merge_fail(
df, 'mean_env',
['sex_id', 'age_group_id', 'year_id', 'location_id'])
for col in self.deaths_cols:
df['cf' + col.split('deaths')[1]] = df[col] / df['mean_env']
df = df.drop('mean_env', axis=1)
return df
def make_age_standardized_group(self, df, age_weight_dict):
df = df.copy()
# get the age weights
df['weight'] = df['age_group_id'].map(age_weight_dict)
report_if_merge_fail(df, 'weight', 'age_group_id')
df = add_population(df, pop_df=self.pop_df)
report_if_merge_fail(
df, "population",
['sex_id', 'age_group_id', 'year_id', 'location_id'])
df = add_envelope(df, env_df=self.env_df)
report_if_merge_fail(
df, "mean_env",
['sex_id', 'age_group_id', 'year_id', 'location_id'])
# age standardized deaths rates = deaths / population * weight
for col in self.cf_cols:
df[col] = (
(df[col] * df['mean_env']) / df['population']) * df['weight']
df['age_group_id'] = 27
df = df.drop(['weight', 'population', 'mean_env'], axis=1)
df = df.groupby(self.id_cols,
as_index=False)[self.cf_cols + ['sample_size']].sum()
report_duplicates(df, self.id_cols)
return df
def read_age_restriction_targets_df(self, code_system_id):
df = (
pd.read_csv(
self.art_path.format(code_system_id=code_system_id)).merge(
self.cause_meta_df[["acause", "cause_id"]],
how="left") # add cause_id for source
.pipe(self.expand_age_logic_to_age_group_id) # add age_group_id
.pipe(self.get_code_id_for_package_name,
code_system_id) # add code_id
.rename(columns={"code_id": "target_code_id"})
# acause is NaN only when there is a GC in target_code_id
.fillna(
{
"target_acause": "_gc",
"is_young_invalid": 0,
"override_cause_hierarchy": 0,
}).merge(
self.cause_meta_df[["acause", "cause_id"]].rename(
columns={
"acause": "target_acause",
"cause_id": "target_cause_id"
}),
how="left",
)) # add cause_id for target
misc.report_if_merge_fail(df, "cause_id", "acause")
misc.report_if_merge_fail(df, "target_cause_id", "target_acause")
return df[[
"cause_id",
"is_young_invalid",
"override_cause_hierarchy",
"age_group_id",
"target_cause_id",
"target_code_id",
]]
def simple_aggregate(self):
"""Aggregate location_ids to country level."""
df = self.df.copy()
country_location_ids = \
get_country_level_location_id(df.location_id.unique(),
self.location_meta_df)
df = df.merge(country_location_ids, how='left', on='location_id')
report_if_merge_fail(df, 'country_location_id', ['location_id'])
df = df[df['location_id'] != df['country_location_id']]
df['location_id'] = df['country_location_id']
df = df.drop(['country_location_id'], axis=1)
# want to collapse site_id for national level
group_cols = [col for col in df.columns if col not in self.val_cols]
group_cols.remove('site_id')
df = df.groupby(group_cols, as_index=False)[self.val_cols].sum()
# set site_id for national aggregates (cannot be missing)
df['site_id'] = 2
# append national aggregates to the incoming dataframe
df = df.append(self.df)
return df