def get_models(regions, targets, duration) -> dict:
models = {}
for region in regions:
with st.spinner(_("Processing {name}").format(name=region.name)):
result = process_region(region, targets, duration)
models.update({(region, k): v for k, v in result.items()})
return models
def __get_models(self) -> dict:
models = {}
regions = self.user_inputs["regions"]
for region in regions:
with st.spinner(_("Processing {name}").format(name=region.name)):
result = self.__process_region(region)
models.update({(region, k): v for k, v in result.items()})
return models
def sari_br_state_dataframe(region: Region) -> pd.DataFrame:
"""
Return the full table of SARI hospital vigilance for the given region.
"""
region = mundi.region(region)
content = sari_br_state_content(region.id)
lines = content.splitlines()
content = lines[0] + b"\n" + b"\n".join(lines[-1000:])
fd = io.BytesIO(content)
with st.spinner(f"Converting to CSV ({region.name})"):
chunks = []
date_columns = [
"dataNotificacao",
"dataInicioSintomas",
"dataNascimento",
"dataEncerramento",
"dataTeste",
]
for df in pd.read_csv(
fd,
index_col=0,
sep=";",
parse_dates=date_columns,
dtype=DTYPES,
converters=CONVERTERS,
engine="c",
chunksize=1000,
encoding="latin1",
):
df: pd.DataFrame = (df.astype(DTYPES).rename(
columns=RENAME).astype({
"status": Status.categories,
"gender": Gender.categories,
"evolution": Evolution.categories,
"test_status": Test.categories,
}))
def localtime(x):
if pd.isna(x):
return x
return x.time()
df["notification_time"] = df["notification_date"].apply(localtime)
df["notification_date"] = df["notification_date"].apply(
lambda x: x if pd.isna(x) else x.date())
df.index.name = "id"
chunks.append(df)
df = pd.concat(chunks)
return df
def get_dataframe(self, days, columns, info_cols=()):
regions = self.user_inputs["regions"]
steps = len(self.user_inputs["regions"])
duration = max(days)
days_ranges = np.array([0, *days])
columns = list(columns)
progress_bar = st.progress(0)
with st.spinner(_("Running simulations")):
rows = {}
for i, region in enumerate(regions):
base, group = self.__run_simulations(region, duration)
progress_bar.progress(int(100 * i / steps))
cols = {}
for a, b in sk.window(2, days_ranges):
day = b
a += base.time + 1
b += a - 1
renames = dict(zip(itertools.count(), columns))
name = _("{} days").format(day)
cols[name] = (
pd.DataFrame(group[columns, a:b].max(0))
.T.rename(columns=renames)
.rename(index={0: region.id})
.astype(int)
)
keys = [*cols]
cols_data = [*cols.values()]
rows[region.id] = pd.concat(cols_data, axis=1, names=[_("days")], keys=keys)
progress_bar.empty()
cols_data = pd.concat(list(rows.values()))
cols_data.index = rows.keys()
if info_cols:
extra_info = cols_data.mundi[info_cols]
extra_info = extra_info.astype(object) # streamlit bug?
extra_info.columns = pd.MultiIndex.from_tuples(("", "info", x) for x in extra_info.columns)
data = pd.concat([extra_info, cols_data], axis=1)
return cols_data.sort_values(cols_data.columns[0])
else:
return cols_data.sort_index()
