def aggregate_injuries(self, df):
ct = dbtrees.causetree(
cause_set_version_id=self.como_version.cause_set_version_id)
df = agg_hierarchy(tree=ct,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
df = df[self.index_cols + self.draw_cols]
# aggregate ncodes
df = df.merge(self.como_version.ncode_hierarchy)
df_agg = df.copy()
df_agg = df_agg.groupby([
"age_group_id", "location_id", "year_id", "sex_id", "measure_id",
"cause_id", "parent_id"
])[self.draw_cols].sum().reset_index()
df_agg = df_agg.rename(columns={"parent_id": "rei_id"})
# set attribute
df = df.append(df_agg)
df = df[self.index_cols + self.draw_cols]
for col in self.index_cols:
df[col] = df[col].astype(int)
return df
def compute_ylds(self):
df = self._NE_ylds
df = df.append(self.injury_inputs.short_term_en_ylds)
# aggregate cases
cause_tree = self.get_cause_tree()
df = agg_hierarchy(tree=cause_tree,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
df = df[self.index_cols + self.draw_cols]
# aggregate ncodes
df = df.merge(self.como_version.ncode_hierarchy)
df_agg = df.copy()
df_agg = df.groupby([
"age_group_id", "location_id", "year_id", "sex_id", "measure_id",
"cause_id", "parent_id"
])[self.draw_cols].sum().reset_index()
df_agg = df_agg.rename(columns={"parent_id": "rei_id"})
# set attribute
df = df.append(df_agg)
df = df[self.index_cols + self.draw_cols]
self.ylds = df
def aggregate_cause(self, df, cause_set_version_id):
ct = dbtrees.causetree(cause_set_version_id=cause_set_version_id)
df = common.agg_hierarchy(tree=ct,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
df = df[self.index_cols + self.draw_cols]
for col in self.index_cols:
df[col] = df[col].astype(int)
all_done = df.cause_id.unique().tolist()
cflat = view_cause_hierarchy_history(
self.como_version.reporting_cause_set_version_id)
cflat = cflat[[
'cause_id', 'path_to_top_parent', 'parent_id', 'cause_name',
'acause', 'cause_outline'
]]
roots = cflat[cflat.cause_id == cflat.parent_id].parent_id
cts = []
for root in roots:
thisflat = cflat[cflat.path_to_top_parent.str.startswith(
str(root))]
ct = tree.parent_child_to_tree(
thisflat,
'parent_id',
'cause_id',
info_cols=['cause_name', 'acause', 'cause_outline'])
cts.append(ct)
reporting_dfs = []
for ct in cts:
rep_only = common.agg_hierarchy(tree=ct,
df=df.copy(deep=True),
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
reporting_dfs.append(rep_only)
reporting_df = pd.concat(reporting_dfs)
reporting_df = reporting_df[~reporting_df["cause_id"].isin(all_done)]
reporting_df = reporting_df[self.index_cols + self.draw_cols]
for col in self.index_cols:
reporting_df[col] = reporting_df[col].astype(int)
return pd.concat([df, reporting_df]).reset_index(drop=True)
def aggregate_causes(self):
ct = dbtrees.causetree(
cause_set_version_id=self.como_version.cause_set_version_id)
ct = deepcopy(ct)
self.imp_df = common.agg_hierarchy(tree=ct,
df=self.imp_df,
index_cols=self.index_cols +
["cause_id", "rei_id"],
data_cols=self.draw_cols,
dimension="cause_id")
def compute_incidence(self):
cause_tree = self.get_cause_tree()
inci = self._most_detailed_incidence.append(
self._short_term_injury_incidence)
inci = common.agg_hierarchy(tree=cause_tree,
df=inci,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
self.incidence = inci
def aggregate_cause(self, df, cause_set_version_id):
ct = dbtrees.causetree(cause_set_version_id=cause_set_version_id)
df = common.agg_hierarchy(tree=ct,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
df = df[self.index_cols + self.draw_cols]
for col in self.index_cols:
df[col] = df[col].astype(int)
return df
def aggregate_sequela(self, df):
seq_tree = dbtrees.sequelatree(
sequela_set_version_id=self.como_version.sequela_set_version_id)
df = df[self.index_cols + self.draw_cols]
df = df.groupby(self.index_cols).sum().reset_index()
df = agg_hierarchy(
tree=seq_tree,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="sequela_id")
df = df[self.index_cols + self.draw_cols]
for col in self.index_cols:
df[col] = df[col].astype(int)
return df
def compute_ylds(self):
df = self._most_detailed_NE_injuries
df = df.append(self._short_term_injury_ylds)
df = df.groupby(self.index_cols).sum().reset_index()
df = df.append(self._get_simulated_ylds())
df = df.groupby(self.index_cols).sum().reset_index()
df = df.append(self._other_drug(df))
df = df.append(self._get_residuals(df))
# aggregate
cause_tree = self.get_cause_tree()
df = common.agg_hierarchy(tree=cause_tree,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
df = df[self.index_cols + self.draw_cols]
self.ylds = df
def _ecode_prevalence(self):
if self.injury_inputs is None:
raise AttributeError(
"can't access _ecode_prevalence without setting injury_inputs")
# get the injuries hierarchy
cause_tree = self.get_cause_tree()
for node in cause_tree.level_n_descendants(1):
if node.id != 687:
cause_tree.prune(node.id)
# aggregate from most detailed
df = self.injury_inputs.ecode_prevalence.copy()
df = common.agg_hierarchy(tree=cause_tree,
df=df,
index_cols=self.index_cols,
data_cols=self.draw_cols,
dimension="cause_id")
df = df[df.cause_id != 294] # drop all cause from injuries
return df[self.index_cols + self.draw_cols]
def compute_aggregates(df, dim, cv):
# grab the cause tree
cause_tree = get_cause_tree(cv)
# aggregate ecodes
df = agg_hierarchy(tree=cause_tree,
df=df,
index_cols=dim.index_names,
data_cols=dim.data_list(),
dimension="cause_id")
df = df[dim.index_names + dim.data_list()]
# aggregate ncodes
df = df.merge(cv.ncode_hierarchy)
df_agg = df.copy()
df_agg = df.groupby([
"age_group_id", "location_id", "year_id", "sex_id", "cause_id",
"parent_id", "measure_id"
])[dim.data_list()].sum().reset_index()
df_agg = df_agg.rename(columns={"parent_id": "rei_id"})
# set attribute
df = df.append(df_agg)
df = df[dim.index_names + dim.data_list()]
return df
