💡 研究假设

import pandas as pd
from scipy import stats

df = pd.read_csv('ncra_master.csv')
corr, p = stats.pearsonr(df['aging_index'].dropna(), df['psychiatrists_per_100k'].dropna())
print(f'r={corr:.3f}, p={p:.4f}')

df <- read.csv('ncra_master.csv')
cor.test(df$aging_index, df$psychiatrists_per_100k, method='pearson')

import pandas as pd
import statsmodels.formula.api as smf

df = pd.read_csv('ncra_master.csv')
model = smf.ols('ncra_risk_score ~ dependency_ratio + C(country_iso)', data=df).fit()
print(model.summary())

df <- read.csv('ncra_master.csv')
library(plm)
pdata <- pdata.frame(df, index=c('country_iso','year'))
model <- plm(ncra_risk_score ~ dependency_ratio, data=pdata, model='within')
summary(model)

import pandas as pd
import statsmodels.formula.api as smf

df = pd.read_csv('ncra_sex_disaggregated.csv')
model = smf.ols('ncra_risk_score ~ sex * gdp_per_capita_usd', data=df).fit()
print(model.params[['sex','gdp_per_capita_usd','sex:gdp_per_capita_usd']])

df <- read.csv('ncra_sex_disaggregated.csv')
model <- lm(ncra_risk_score ~ sex * gdp_per_capita_usd, data=df)
summary(model)

import pandas as pd

df = pd.read_csv('ncra_master.csv').sort_values(['country_iso','year'])
df['ddi_lag5'] = df.groupby('country_iso')['demographic_dividend_index'].shift(5)
from scipy import stats
corr, p = stats.pearsonr(df['ddi_lag5'].dropna(), df['gdp_per_capita_usd'].dropna())
print(f'Lagged correlation: r={corr:.3f}, p={p:.4f}')

df <- read.csv('ncra_master.csv')
df <- df[order(df$country_iso, df$year),]
df$ddi_lag5 <- ave(df$demographic_dividend_index, df$country_iso, FUN=function(x) c(rep(NA,5),head(x,-5)))
cor.test(df$ddi_lag5, df$gdp_per_capita_usd, use='complete.obs')

import pandas as pd
import numpy as np

df = pd.read_csv('ncra_master.csv')
df['post_covid'] = (df['year'] >= 2020).astype(int)
df['time_since_covid'] = np.maximum(0, df['year'] - 2020)
from statsmodels.formula.api import ols
model = ols('aging_index ~ year + post_covid + time_since_covid + C(country_iso)', data=df).fit()
print(model.params[['post_covid','time_since_covid']])

df <- read.csv('ncra_master.csv')
df$post_covid <- as.integer(df$year >= 2020)
df$time_since_covid <- pmax(0, df$year - 2020)
model <- lm(aging_index ~ year + post_covid + time_since_covid + country_iso, data=df)
summary(model)