def decline_50_plus(ref, past_pop):
loc_list = list(db.get_locations_by_level(3).location_id)
ref_2017 = past_pop.sel(
year_id=2017, age_group_id=22, location_id=loc_list, sex_id=3) / 1e9
ref_2100 = ref.sel(
year_id=2100, age_group_id=22, location_id=loc_list, sex_id=3).sel(
scenario=0, drop=True) / 1e9
change = (ref_2100 - ref_2017) / ref_2017
mean_change = change.mean('draw')
fifty_plus = change.mean('draw').where(
change.mean('draw') < -0.5).dropna("location_id")
china = change.sel(location_id=6).mean('draw')
japan = change.sel(location_id=67).mean('draw')
twentyfive_to_fifty = mean_change\
.where(mean_change < -0.25).where(mean_change > -0.50)\
.dropna("location_id")
ref_slice = past_pop.sel(location_id=fifty_plus.location_id.values,
age_group_id=22,
sex_id=3,
year_id=2017)
print(f"Number of countries decline greater than 50%: {len(fifty_plus)}"\
f"China percent decline: {china.values.round(2)}"\
f"Japan percent decline: {japan.values.round(2)}"\
f"Number of countries decline between 25% - 50%: "\
f"{len(twentyfive_to_fifty)}")
return change
def super_region_pop(pop_da):
locs = db.get_locations_by_level(1)[['location_id', 'location_name']]
supreg_pop = pop_da.sel(location_id=locs.location_id.tolist())
supreg_df = supreg_pop.sel(quantile="mean").drop(
["scenario", "sex_id", "age_group_id"]).to_dataframe().reset_index()
peak_vals = supreg_df.groupby("location_id")["population"].max()
peaks = supreg_df.merge(peak_vals).merge(locs)
peak_years = peaks.year_id
if (peak_years < 2100).all() != True:
warnings.warn("Not all super regions peak!")
supreg_17_100 = supreg_df[supreg_df.year_id.isin([2017, 2100])].drop(
"quantile", axis=1).set_index(
["location_id",
"year_id"]).unstack("year_id").droplevel(0, axis=1).reset_index()
supreg_17_100["diff"] = supreg_17_100[2100] - supreg_17_100[2017]
pos_diff_locs = supreg_17_100[
supreg_17_100["diff"] > 0].location_id.tolist()
supreg_higher100 = supreg_pop.sel(year_id=2100, location_id=pos_diff_locs)
supreg_higher100_mean_ui = supreg_higher100.drop([
"scenario", "sex_id", "age_group_id", "year_id"
]).to_dataframe().unstack().droplevel(0, axis=1).reset_index().merge(locs)
print("SUPER-REGIONS WITH HIGHER POP IN 2100 THAN 2017 (IN 2100)")
print(supreg_higher100_mean_ui)
print("\nREMINDER: check plots for most severe declines\n")
def super_region_replacement(tfr_da):
locs = db.get_locations_by_level(1)[['location_id', 'location_name']]
tfr_df = tfr_da.drop("scenario").to_dataframe().reset_index()
tfr_wide = tfr_df.set_index(
["location_id", "year_id", "quantile"]).unstack().droplevel(
level=0, axis=1).reset_index()
supreg_df = tfr_wide.merge(locs, on="location_id")
below_rep_17 = supreg_df.query("year_id==2017 & mean<2.1").round(N_DECIMALS)
below_rep_17_names = "; ".join(below_rep_17.location_name.unique())
num_below_17 = len(below_rep_17.location_id.unique())
supreg_100 = supreg_df.query("year_id==2100").round(N_DECIMALS)
supreg_100_min = supreg_100[supreg_100["mean"]==supreg_100["mean"].min()]
supreg_100_max = supreg_100[supreg_100["mean"]==supreg_100["mean"].max()]
subsah = supreg_df.query("location_name=='Sub-Saharan Africa'").round(
N_DECIMALS)
subsah_17 = subsah.query("year_id==2017")
subsah_100 = subsah.query("year_id==2100")
subsah_1st_yr_below_rep = subsah.query("mean<2.1").year_id.min()
print(
f"{num_below_17} GBD super-regions had reached a below-replacement TFR "
f"(<2·1) by 2017: {below_rep_17_names}. By 2100, our reference TFR "
f"forecasts for the super-regions were between "
f"{supreg_100_min['mean'].values[0]} (95% UI "
f"{supreg_100_min.lower.values[0]}–{supreg_100_min.upper.values[0]}) "
f"for {supreg_100_min.location_name.values[0]} "
f"and {supreg_100_max['mean'].values[0]} ({supreg_100_max.lower.values[0]}– "
f"{supreg_100_max.upper.values[0]}) for "
f"{supreg_100_max.location_name.values[0]} "
"(table 1). Sub-Saharan Africa is forecasted to decline from "
f"{subsah_17['mean'].values[0]} in 2017 to {subsah_100['mean'].values[0]} "
f"({subsah_100.lower.values[0]}–{subsah_100.upper.values[0]}) "
f"in 2100 and reach below-replacement in {subsah_1st_yr_below_rep}.\n"
)
from fbd_core.etl import resample
from fbd_core.etl.aggregator import Aggregator
from fbd_core.etl.transformation import expand_dimensions
from fbd_core.file_interface import FBDPath, open_xr, save_xr
import sys
sys.path.append(".")
import settings
ALL_AGE_ID = 22
BOTH_SEX_ID = 3
N_DECIMALS = 2
COUNTRIES = db.get_locations_by_level(3).location_id.values.tolist()
def _helper_round(val, divide_by_this):
return round(val/divide_by_this, N_DECIMALS)
def _location_id_to_name(location_ids):
locs = db.get_locations_by_max_level(3)[['location_id', 'location_name']]
locs = locs[locs['location_id'].isin(location_ids)]
location_names = locs['location_name'].tolist()
return location_names
# The global total fertility rate (TFR) in 2100 is forecasted to be 1.66
# (95% UI xx–xx).
def tfr_subsaharan_africa(tfr_da):
locs = db.get_locations_by_level(3)[['location_id', 'location_name',
'super_region_name']]
subsah_countries = locs.query(
"super_region_name=='Sub-Saharan Africa'").location_id.tolist()
subsah_100 = tfr_da.sel(location_id=subsah_countries, year_id=2100).drop(
["scenario", "year_id"])
# specific countries
nga_100 = subsah_100.sel(location_id=214) # nigeria
nga_100_mean, nga_100_lower, nga_100_upper = get_quantiles_and_round(
nga_100, N_DECIMALS)
ner_100 = subsah_100.sel(location_id=213) # niger
ner_100_mean, ner_100_lower, ner_100_upper = get_quantiles_and_round(
ner_100, N_DECIMALS)
sa_100 = subsah_100.sel(location_id=196) # south africa
sa_100_mean, sa_100_lower, sa_100_upper = get_quantiles_and_round(
sa_100, N_DECIMALS)
# countries above replacement in 2100
above_rep_100 = subsah_100.round(N_DECIMALS).to_dataframe().unstack().\
transpose().reset_index().query("mean>2.1").merge(locs).\
sort_values(by="mean", axis=0, ascending=False).reset_index(drop=True)
abvrep_rows = [] # subset each rows
for row in range(0, len(above_rep_100)):
abvrep_rows.append(above_rep_100.iloc[row])
abvrep_strings = [] # turn each row into a string
for row in range(0, len(abvrep_rows)):
if row == 0:
string = (f"{abvrep_rows[0].location_name} with a TFR of "
f"{abvrep_rows[0]['mean']} ({abvrep_rows[0].lower}–"
f"{abvrep_rows[0].upper})")
else:
string = (f"{abvrep_rows[row].location_name} "
f"({abvrep_rows[row]['mean']} [{abvrep_rows[row].lower}–"
f"{abvrep_rows[row].upper}])")
abvrep_strings.append(string) # make list of string
abvrep_string = ", ".join(abvrep_strings) # join list to make final string
# lowest 5 tfr countries
lowest_5_100 = subsah_100.to_dataframe().unstack().transpose().reset_index().\
nsmallest(5, "mean").reset_index().merge(locs).round(N_DECIMALS)
# names ascending order
lowest_ascending = ", ".join(lowest_5_100.location_name)
# get range
lowest_low = (f"{lowest_5_100.iloc[0]['mean']} ({lowest_5_100.iloc[0].lower}–"
f"{lowest_5_100.iloc[0].upper})")
highest_low = (f"{lowest_5_100.iloc[4]['mean']} ({lowest_5_100.iloc[4].lower}–"
f"{lowest_5_100.iloc[4].upper})")
print(
"In 2017, estimates of TFR were 5·11, 7·08, and 2·29 in Nigeria, Niger, "
"and South Africa, respectively; by 2100, our reference forecasts were "
f"{nga_100_mean} (95% UI {nga_100_lower}–{nga_100_upper}), {ner_100_mean} "
f"({ner_100_lower}–{ner_100_upper}), and {sa_100_mean} ({sa_100_lower}–"
f"{sa_100_upper}) (table 1). Only four countries in sub-Saharan Africa "
f"are forecasted to stay above replacement in 2100: {abvrep_string}. "
f"Lowest TFR reference forecasts for sub-Saharan Africa in 2100 were for "
f"{lowest_ascending}, all between {lowest_low} and {highest_low}.\n"
)
def pop_largest_countries(pop_da):
locs_w_supreg = db.get_locations_by_level(3)[[
'location_id', 'location_name', "super_region_name"
]]
locs = locs_w_supreg.drop("super_region_name", axis=1)
largest_5_2100 = pop_da.sel(year_id=2100, location_id=COUNTRIES).drop(
["year_id", "age_group_id", "sex_id",
"scenario"]).to_dataframe().unstack("quantile").droplevel(
0, axis=1).reset_index().nlargest(5, "mean").merge(locs)
print("LARGEST 5 COUNTRIES IN 2100")
print(largest_5_2100)
print("\nCHECK VIA PLOTS:\n"
"'Nigeria is forecasted with continued growth through 2100 "
"and is expected to be the 2nd most populous country by then. The "
"reference forecasts for China and India peak well before 2050 and "
"both countries thereafter follow steep declining trajectories...'")
chn_peak_year = pop_da.sel(location_id=6, quantile="mean")
chn_peak_year = chn_peak_year.where(chn_peak_year == chn_peak_year.max(),
drop=True).year_id.values[0]
chn_peak_yr_da = pop_da.sel(location_id=6,
year_id=chn_peak_year).drop("year_id")
ind_peak_year = pop_da.sel(location_id=163, quantile="mean")
ind_peak_year = ind_peak_year.where(ind_peak_year == ind_peak_year.max(),
drop=True).year_id.values[0]
ind_peak_yr_da = pop_da.sel(location_id=163,
year_id=ind_peak_year).drop("year_id")
usa_peak_year = pop_da.sel(location_id=102, quantile="mean")
usa_peak_year = usa_peak_year.where(usa_peak_year == usa_peak_year.max(),
drop=True).year_id.values[0]
usa_peak_yr_da = pop_da.sel(location_id=102,
year_id=usa_peak_year).drop("year_id")
chn_ind_usa_peak = xr.concat(
[chn_peak_yr_da, ind_peak_yr_da, usa_peak_yr_da], dim="location_id")
chn_ind_usa_100 = pop_da.sel(year_id=2100, location_id=[6, 163, 102])
pcnt_2100_to_peak_chn_ind_usa = (
(chn_ind_usa_100 / chn_ind_usa_peak) * 100).round(N_DECIMALS).drop(
["year_id", "age_group_id", "sex_id",
"scenario"]).to_dataframe().unstack("quantile").droplevel(
0, axis=1).reset_index().merge(locs)
print("\nCHINA INDIA USA PCT OF PEAK IN 2100")
print(pcnt_2100_to_peak_chn_ind_usa)
print(
"\nCHECK WITH TABLES BELOW:\n"
"'The USA is projected to experience population growth until "
"mid-century, followed by only moderate decline of less than 10% by "
"2100. We forecast that the number of countries in sub-Saharan Africa "
"among the countries with the 10 largest populations will increase "
"from only Nigeria in 2017 to also include DR Congo, Ethiopia, and "
"Niger in 2100.'")
largest_10_2017 = pop_da.sel(year_id=2017, location_id=COUNTRIES).drop([
"year_id", "age_group_id", "sex_id", "scenario"
]).to_dataframe().unstack("quantile").droplevel(
0,
axis=1).reset_index().nlargest(10, "mean").merge(locs_w_supreg).drop(
["lower", "upper"],
axis=1).query("super_region_name == 'Sub-Saharan Africa'")
print("\nSUB-SAHARAN COUNTRIES FROM 10 LARGEST COUNTRIES 2017")
print(largest_10_2017)
largest_10_2100 = pop_da.sel(year_id=2100, location_id=COUNTRIES).drop([
"year_id", "age_group_id", "sex_id", "scenario"
]).to_dataframe().unstack("quantile").droplevel(
0,
axis=1).reset_index().nlargest(10, "mean").merge(locs_w_supreg).drop(
["lower", "upper"],
axis=1).query("super_region_name == 'Sub-Saharan Africa'")
print("\nSUB-SAHARAN COUNTRIES FROM 10 LARGEST COUNTRIES 2100")
print(largest_10_2100)
