R/CurrencyConversion.R

In l-gorman/rhomis-R-package: Essential Processes for RHoMIS Data

#' Query World Bank Statistics
#'
#' Find out PPP conversions based on the
#' world bank statistics
#'
#' Rpackage file: CurrencyConversion.R
#'
#' @param year The year you are hoping to query
#' @param country_code The two letter ISO country code
#'
#' @return
#' @export
#'
#' @examples
querywb_stats <- function(year, country_code) {
    tryCatch(
        {
            wbstats::wb_data(
                indicator = "PA.NUS.PRVT.PP",
                country = country_code,
                start_date = year
            )

        },
        error = function(cond) {
            message("Could not obtain currency conversion factor")
            message("error message:")
            message(cond)
            # Choose a return value in case of error
            return(tibble::as_tibble(list(
                "PA.NUS.PRVT.PP" = c(NA),
                "date" = c(NA)
            )))
        },
        finally = {

        }
    )
}


#' Conversion_factor
#'
#' Identify the closest available currency conversion factor
#' based on world bank data. If we cannot find the current
#' currency conversion factor, we will check for the most recent,
#' going as far back as 10 years
#'
#' Rpackage file: CurrencyConversion.R
#'
#' @param year The year you are querying the conversion factors for
#' @param country_code The country codes that you are querying for
#'
#' @return
#' @export
#'
#' @examples
currency_conversion_factor <- function(year, country_code) {
    conversion_factor <- NA
    year <- as.numeric(year)
    country_code <- as.character(country_code)
    initial_year <- year

    if (is.na(year) | is.na(country_code)) {
        message("Either country or year of survey are not well defined, so unable to obtain currency conversion")

        return(list(
            "conversion_year" = NA,
            "conversion_factor" = NA
        ))
    }

    # Deciding whether or not to query
    # World bank stats

    max_year <- max(currency_conversion$year)

    if (year>max_year){
        use_query <- T
    }else{
        use_query <- F
    }


    #Fetch using inbuilt conversion factors
    if(use_query==F){

        three_letter_code <- tolower(country_codes[["alpha-3"]])[tolower(country_codes[["alpha-2"]])==tolower(country_code)]
        three_letter_code <- three_letter_code[!is.na(three_letter_code)]

        if (length(three_letter_code)!=1)
        return(list(
            "conversion_year" = NA,
            "conversion_factor" = NA
        ))

        while (is.na(conversion_factor) & initial_year - year < 10) {

            row_subset <- tolower(currency_conversion[["Country Code"]])==three_letter_code &
                as.numeric(currency_conversion[["year"]])==as.numeric(year)


            conversion_factor <- as.numeric(currency_conversion[row_subset,"value"])
            conversion_year <- as.numeric(year)

            if (length(conversion_factor)!=1){
                conversion_factor <- NA
                conversion_year <- NA
            }

            if (is.na(conversion_factor)) {
                conversion_year <- NA
            }

        }



    }



    # Fetch by querying WB database
    if(use_query==T){

        while (is.na(conversion_factor) & initial_year - year < 10) {
            Sys.sleep(0.5)

            wb_result <-querywb_stats(year, country_code)


            conversion_factor <- as.numeric(wb_result$PA.NUS.PRVT.PP)
            conversion_year <- as.numeric(wb_result$date)

            if (nrow(wb_result) == 0) {
                conversion_factor <- NA
                conversion_year <- NA
            }


            year <- year - 1

            if (is.na(conversion_factor)) {
                conversion_year <- NA
            }
        }
    }

    return(list(
        "conversion_year" = conversion_year,
        "conversion_factor" = conversion_factor
    ))
}


#' Convert all Currencies
#'
#' A function for finding the relevant currency conversion
#' factors for all household interviews
#'
#' Rpackage file: CurrencyConversion.R
#'
#' @param data The data containing a column of ISO country codes and
#' a column of Years
#' @param country_column The name of the country column
#' @param year_column The name of the year column
#'
#' @return
#' @export
#'
#' @examples
convert_all_currencies <- function(data, country_column = "country", year_column = "year") {

    # data <- tibble::as_tibble((list(
    #
    #     "country_code"=c("KM","KM","VN", "VN", "UG", NA, "XY"),
    #     "year"=c("2016","2016","2021", "2014", "2016", NA, "2020")
    # )))
    subset_data <- data %>% dplyr::select(c(country_column, year_column))


    combinations <- tibble::as_tibble(table(subset_data,useNA="always")) %>%
        dplyr::filter(n > 0) %>%
        dplyr::select(-c("n"))

    conversion_factors <- c()
    conversion_years <- c()

    for (i in 1:nrow(combinations)) {
        conversion_data <- currency_conversion_factor(year=combinations[i, year_column], country_code=combinations[i, country_column])
        conversion_factors <- c(conversion_factors, conversion_data[["conversion_factor"]])
        conversion_years <- c(conversion_years, conversion_data[["conversion_year"]])
    }

    combinations$conversion_factor <- unlist(conversion_factors)
    combinations$conversion_year <- unlist(conversion_years)

    matching_list <- c(country_column, year_column)
    names(matching_list) <- c(country_column, year_column)

    data[[year_column]] <- as.numeric(data[[year_column]])
    combinations$year <- as.numeric(combinations$year)

    combinations <- combinations[!duplicated(combinations),]

    data_with_conversions <- dplyr::left_join(data, combinations, by = matching_list,)

    data <- add_column_after_specific_column(
        data = data,
        new_data = data_with_conversions["conversion_factor"],
        new_column_name = "conversion_factor",
        old_column_name = year_column,
        loop_structure = F
    )

    data <- add_column_after_specific_column(
        data = data,
        new_data = data_with_conversions["conversion_year"],
        new_column_name = "conversion_year",
        old_column_name = "conversion_factor",
        loop_structure = F
    )

    return(data)
}