periods.R

pacman::p_load(tidyverse, rio, magrittr, janitor)
# This is a function for dividing a time series into periods, with allowence for gaps (threshold)
# data = dataset
# variable = time for measurement, e.g. year (duplicates allowed)
# threshold = number of consecutive years (e.g.) allowed missing in a period
# filter_less_than = filters periods with less than N values (not the length from min to max!)

periods <- function(data,
                    variable,
                    threshold = 3,
                    filter_less_than = FALSE){
  enquo(variable) -> variable_enquo
  data %>% group_vars() %>% syms() -> groups_vec
  out <- data %>%
    dplyr::arrange(!!variable_enquo) %>%
    mutate(period = cumsum(c(1, diff(!!variable_enquo) >= threshold+1))) %>%
    group_by(period, .add=T) %>%
    mutate(n_year_period = n_distinct(!!variable_enquo))
  
  if(!is_empty(groups_vec)){out %>% group_by(!!!groups_vec) -> out}else{out %>% ungroup() -> out}
  
  if(filter_less_than){out %>%
    filter(n_year_period >= filter_less_than) %>%
    mutate(period = cumsum(c(1, diff(!!variable_enquo) >= threshold+1))) ->
      out
  }
  
  return(out)
}