Converts the Cumulative Distribution Function (CDF) of ever-smoking into the Probability Density Function (PDF), which represents the probability of initiating smoking at a specific age.
Usage
p_dense(
data,
cum_func_var,
strat_vars = c("cohort", "sex", "imd_quintile"),
lowest_year = 2003,
max_year = 2100,
cdf_smooth_df = 6
)Arguments
- data
Data table with cumulative probabilities.
- cum_func_var
Character - name of cumulative variable.
- strat_vars
Character vector - stratification variables.
- lowest_year
integer - start year filter.
- max_year
integer - end year filter.
- cdf_smooth_df
Numeric - degrees of freedom for the monotone smooth of the cumulative curve over age. NULL or 0 skips the smooth and restores the pre-2026 behaviour.
Details
The cumulative curve is smoothed over age within each group before it is differenced (see smooth_cdf). Set cdf_smooth_df = NULL to skip that and get the old behaviour back, which is useful for comparing runs, but note that the old behaviour is chaotic: see the note on blank_zeros in p_smooth.