R/est_trend.R

In nrennie/WBLifeExpectancy: Life Expectancy Analysis of World Bank data

#' Estimate Trend
#'
#' This function estimates trend in life expectancy over a moving window.
#'
#' @param input_vec A numeric vector of life expectancies across different years.
#' @param year_vec A numeric vector of the years observed. Default 1960:2020,
#' @param window_val A numeric giving length of window. Default 10.
#' @return A numeric vector.
#' @export


est_trend <- function(input_vec,
                      year_vec = 1960:2020,
                      window_val = 10) {
  #check conditions
  if (length(input_vec) != length(year_vec)) {
    stop("input_vec and year_vec must be the same length")
  }

  #identify data range
  test_years <- (year_vec[1] + window_val - 1) : utils::tail(year_vec, 1)

  #empty vectors to store data
  est_beta <- numeric(length(test_years))
  upper_beta <- numeric(length(test_years))
  lower_beta <- numeric(length(test_years))

  #apply linear models
  for (i in seq_len(length(test_years))) {
    test_year <- test_years[i]
    #select data
    x <- year_vec[(which(year_vec == test_year - window_val + 1)):which(year_vec == test_year)] # nolint
    y <- input_vec[(which(year_vec == test_year - window_val + 1)):which(year_vec == test_year)] # nolint

    #fit model
    mod <- stats::lm(y ~ x)
    est_beta[i] <- mod$coefficients[2]

    #intervals
    k <- stats::confint(mod)
    upper_beta[i] <- k[2, 2]
    lower_beta[i] <- k[2, 1]
  }

  #output values
  list("Trend estimate" = est_beta,
       "Lower limit" = lower_beta,
       "Upper limit" = upper_beta,
       "Test years" = test_years)
}