Super slow

[johanneskoch94]

This should be replaced with a pre-filled out source, so that it doesn't have to be done for every conversion.

GDPuc/R/adapt_source.R

Lines 94 to 150 in f107ebc

	if ("with_USA" %in% replace_NAs) {
	USA_def_growth <- source %>%
	dplyr::filter(.data$iso3c == "USA") %>%
	dplyr::select("year", "gd" = "GDP deflator") %>%
	dplyr::mutate("gd" = .data$gd / dplyr::lag(.data$gd))
	source_adapted <- source_adapted %>%
	dplyr::filter(.data$iso3c %in% unique(gdp$iso3c)) %>%
	# Fill in the MER and PPPs with the growth rates from the USA (= 1)
	dplyr::group_by(.data$iso3c) %>%
	tidyr::fill(c("MER (LCU per US$)",
	"PPP conversion factor, GDP (LCU per international $)"),
	.direction = "downup") %>%
	dplyr::ungroup() %>%
	# For the deflator, we need to multiply the bordering values with the actual USA growth
	dplyr::left_join(USA_def_growth, by = dplyr::join_by("year")) %>%
	# Forward
	dplyr::mutate(
	`GDP deflator` = purrr::accumulate(
	dplyr::row_number(),
	~ dplyr::coalesce(.data$`GDP deflator`[.y], .x * .data$gd[.y]),
	.init = NA
	)[-1],
	.by = c("iso3c")
	) %>%
	# Backward
	dplyr::arrange(-.data$year) %>%
	dplyr::mutate(
	`GDP deflator` = purrr::accumulate(
	dplyr::row_number(),
	~ dplyr::coalesce(.data$`GDP deflator`[.y], .x / .data$gd[.y]),
	.init = NA
	)[-1],
	.by = c("iso3c")
	) %>%
	dplyr::arrange(.data$iso3c, .data$year) %>%
	dplyr::select(-"gd")
	# If there is no data whatsoever for the country, use US values
	ec <- dplyr::group_by(source_adapted, .data$iso3c) %>%
	dplyr::filter(all(is.na(.data$`GDP deflator`))) %>%
	dplyr::pull("iso3c") %>%
	unique()
	source_ec <- source %>%
	dplyr::filter(.data$iso3c == "USA") %>%
	tidyr::complete(iso3c = ec,
	tidyr::nesting(!!!(rlang::syms(colnames(source)[-1])))) %>%
	dplyr::filter(.data$iso3c %in% ec)
	source_adapted <- source_adapted %>%
	dplyr::filter(!.data$iso3c %in% ec) %>%
	dplyr::bind_rows(source_ec)
	}
	source_adapted
	}