不雅的解决方案：

# add to the dataframe a list of dates for the start of each year
df= rbind(df, data.table(date = as.Date(ISOdate(seq(2005,2023),1,1)), "interest_rate" = NA), fill=T)

# set order
setorder(df, -date)

# set value for the new dates to the current interest rate
df[is.na(interest_rate), interest_rate := df[df[is.na(interest_rate)], on = .(date), interest_rate, roll=Inf]]

# identify the previous shift date was
df[, change_date := shift(date, 1L)]

# get next rate change or end of year
df[, next_change_date_or_year := min(as.Date(ISOdate(year,12,31)), change_date, na.rm=T), date]

# get relative duration within year
df[, relative_duration := as.numeric((next_change_date_or_year-date)/365)]

df[, year := year(date)]

# weighted average per year
yearly_IR = df[!is.na(year), .(weight_interest_rate = weighted.mean(`interest rate`, relative_duration)), year]

一种方法是为整个系列中的每一天创建一个利率数据框架，按年份分组，然后对一年中每个日期的利率求平均值。
重要的第一步是将字符串形式的日期转换为实际日期。

library(tidyverse)

df %>%
  mutate(date = as.Date(date, '%d/%m/%Y'),
         days = c(as.numeric(diff(date)), 0)) %>%
  rowwise() %>%
  reframe(date = seq(date, by = 'day', length.out = days),
          interest_rate = interest_rate,
          year = lubridate::year(date)) %>%
  group_by(year) %>%
  summarize(interest_rate = mean(interest_rate))
#> # A tibble: 18 x 2
#>     year interest_rate
#>    <dbl>         <dbl>
#>  1  2006         4.84 
#>  2  2007         5.51 
#>  3  2008         4.67 
#>  4  2009         0.644
#>  5  2010         0.5  
#>  6  2011         0.5  
#>  7  2012         0.5  
#>  8  2013         0.5  
#>  9  2014         0.5  
#> 10  2015         0.5  
#> 11  2016         0.398
#> 12  2017         0.291
#> 13  2018         0.604
#> 14  2019         0.75 
#> 15  2020         0.228
#> 16  2021         0.107
#> 17  2022         1.47 
#> 18  2023         3.97