def _most_detailed_NE_injuries(self):
nemat = self.injury_inputs.en_matrices
nemat = nemat.merge(self.como_sim.ylds,
on=[
"sequela_id", "age_group_id", "location_id",
"year_id", "sex_id"
])
nemat = nemat.join(
pd.DataFrame(data=(nemat.filter(regex="draw.*x").values *
nemat.filter(regex="draw.*y").values),
index=nemat.index,
columns=self.draw_cols))
nemat = common.apply_restrictions(self.como_version.cause_restrictions,
nemat, self.dimensions.data_list())
parallelism = ["location_id", "year_id", "sex_id"]
for slices in self.dimensions.index_slices(parallelism):
filename = "{yid}_{sid}.csv".format(yid=slices[1], sid=slices[2])
directory = os.path.join("FILEPATH", str(slices[0]))
try:
os.makedirs(directory)
except OSError as exception:
if exception.errno != errno.EEXIST:
raise
filepath = os.path.join(directory, filename)
out_df = nemat.loc[(nemat.location_id == slices[0])
& (nemat.year_id == slices[1]) &
(nemat.sex_id == slices[2])]
out_df.to_csv(filepath, index=False)
nemat = nemat[self.index_cols + self.draw_cols]
nemat = nemat.groupby(self.index_cols).sum().reset_index()
return nemat[self.index_cols + self.draw_cols]
def _get_sequela_measure(self, modelable_entity_reader, measure_id):
if self.dimensions is None or self.como_version is None:
raise AttributeError(
"cannot set sequela via this method if como_version is None or"
" dimensions is None")
df_list = []
for key in self.sequela_set:
maybe_df = modelable_entity_reader.reader_results[key]
try:
df = maybe_df[maybe_df.measure_id == measure_id]
except AttributeError:
raise maybe_df
df_list.append(df)
df = pd.concat(df_list)
# attach sequela ids
df = df.merge(self.como_version.sequela_list,
on='modelable_entity_id',
how='left')
# restrict
df = apply_restrictions(self.como_version.cause_restrictions, df,
self.dimensions.data_list())
df = df[['sequela_id', 'cause_id', 'healthstate_id'] +
self.dimensions.index_names + self.dimensions.data_list()]
return df
def attach_sequela_metadata_and_restict(df, ds):
if df.empty:
return df
sequela_list = ds.params["sequela_list"]
dimensions = ds.params["dimensions"]
cause_restrictions = ds.params["cause_restrictions"]
df = df.merge(sequela_list, on='modelable_entity_id', how='left')
df = apply_restrictions(cause_restrictions, df, dimensions.data_list())
return df[['sequela_id', 'cause_id', 'healthstate_id'] +
dimensions.index_names + dimensions.data_list()]
def attach_sexual_violence_metadata_and_restrict(df, ds):
if df.empty:
return df
# cause_id 941 - Sexual violence
df["cause_id"] = 941
dimensions = ds.params["dimensions"]
cause_restrictions = ds.params["cause_restrictions"]
df = apply_restrictions(cause_restrictions, df, dimensions.data_list())
df = df[['cause_id'] + dimensions.index_names + dimensions.data_list()]
df = df.groupby(['cause_id'] + dimensions.index_names).sum()
return df.reset_index()
def attach_injuries_metadata_and_restict(df, ds):
if df.empty:
return df
injury_sequela = ds.params["injury_sequela"]
dimensions = ds.params["dimensions"]
cause_restrictions = ds.params["cause_restrictions"]
df = df.merge(injury_sequela, on='modelable_entity_id', how='left')
df = apply_restrictions(cause_restrictions, df, dimensions.data_list())
return df[[
"modelable_entity_id", "cause_id", "rei_id", "sequela_id",
"healthstate_id"
] + dimensions.index_names + dimensions.data_list()]
def _NE_ylds(self):
nemat = self.injury_inputs.en_matrices
nemat = nemat.merge(self.como_sim.ylds,
on=[
"sequela_id", "age_group_id", "location_id",
"year_id", "sex_id"
])
nemat = nemat.join(
pd.DataFrame(data=(nemat.filter(regex="draw.*x").values *
nemat.filter(regex="draw.*y").values),
index=nemat.index,
columns=self.draw_cols))
nemat = apply_restrictions(self.como_version.cause_restrictions, nemat,
self.dimensions.data_list())
nemat = nemat.merge(self.como_version.injury_dws_by_sequela)
nemat = nemat.merge(self.como_version.ncode_hierarchy,
left_on="n_code",
right_on="rei")
return nemat[self.index_cols + self.draw_cols]
def _compute_en_ylds(self, simulated_ylds, long_term_en_prev,
short_term_en_ylds):
# compute the en matrix
sim_dim = self.dimensions.get_simulation_dimensions(
measures.PREVALENCE)
denom = self._prepare_ncode_aggregates(long_term_en_prev.copy())
denom = denom[sim_dim.index_names + ["sequela_id"] +
sim_dim.data_list()]
numer = long_term_en_prev[sim_dim.index_names +
["cause_id", "sequela_id"] +
sim_dim.data_list()]
en_mat = broadcast(broadcast_onto_df=numer,
broadcast_df=denom,
index_cols=sim_dim.index_names + ["sequela_id"],
operator="/")
en_mat.fillna(0, inplace=True)
en_mat["measure_id"] = measures.YLD
# apply the matrix to the ylds
en_mat = en_mat.merge(simulated_ylds,
on=sim_dim.index_names + ["sequela_id"])
en_mat = en_mat.join(
pd.DataFrame(data=(en_mat.filter(regex="draw.*x").values *
en_mat.filter(regex="draw.*y").values),
index=en_mat.index,
columns=sim_dim.data_list()))
en_mat = apply_restrictions(self.como_version.cause_restrictions,
en_mat, sim_dim.data_list())
# clean the short term ylds
rei_map = self.como_version.injury_sequela[[
"cause_id", "sequela_id", "rei_id"
]]
en_mat = en_mat.merge(rei_map, on=["sequela_id", "cause_id"])
en_mat = en_mat[sim_dim.index_names + ["cause_id", "rei_id"] +
sim_dim.data_list()]
en_mat.fillna(0, inplace=True)
# compute the combine ylds
en_ylds = pd.concat([en_mat, short_term_en_ylds])
en_ylds = en_ylds.groupby(sim_dim.index_names + ["cause_id", "rei_id"])
en_ylds = en_ylds.sum()[sim_dim.data_list()].reset_index()
return en_ylds
def get_short_term_EN_ylds(self):
if self.dimensions is None or self.como_version is None:
raise AttributeError(
"cannot get en ylds via this method if como_version is None or"
" dimensions is None")
en_dimensions = deepcopy(self.dimensions)
en_dimensions.index_dim.replace_level("measure_id", 3)
en_dimensions.index_dim.add_level(
"cause_id",
self.como_version.cause_restrictions.cause_id.unique().tolist())
en_dimensions.index_dim.add_level(
"rei_id",
self.como_version.ncode_hierarchy.rei_id.unique().tolist())
estim_df = self._get_short_term_EN_estimation(en_dimensions)
annual_df = self._get_short_term_EN_annual(en_dimensions)
df = estim_df.append(annual_df)
df = apply_restrictions(self.como_version.cause_restrictions, df,
self.dimensions.data_list())
self.short_term_en_ylds = df
def set_ecode_prevalence_via_me_reader(self, modelable_entity_reader):
df = self._get_ecode_measure(modelable_entity_reader, 5)
df = apply_restrictions(self.como_version.cause_restrictions, df,
self.dimensions.data_list())
self.ecode_prevalence = df
def set_short_term_ecode_incidence_via_me_reader(self,
modelable_entity_reader):
df = self._get_short_term_ecode_measure(modelable_entity_reader, 6)
df = apply_restrictions(self.como_version.cause_restrictions, df,
self.dimensions.data_list())
self.short_term_ecode_incidence = df
def set_short_term_ecode_ylds_via_me_reader(self, modelable_entity_reader):
df = self._get_short_term_ecode_measure(modelable_entity_reader, 3)
df = transform_metric(df, 3, 1)
df = apply_restrictions(self.como_version.cause_restrictions, df,
self.dimensions.data_list())
self.short_term_ecode_ylds = df