R/match-records.R
In caseybreen/wcensoc:
Defines functions
append_A_to_B
preprocess_B
preprocess_A
#########################################
# created by UGUR YILDIRIM (2020-09-30) #
# revised by WON-TAK JOO (2022-11-10) #
#########################################
# Define %>%
`%>%` <- magrittr::`%>%`
# Define preprocess_A
#' @export
preprocess_A <- function(file_A, fname_var, lname_var, time_var, id_var) {
# Keep fname_var, lname_var, time_var, id_var
file_A <- subset(file_A, select=c(fname_var, lname_var, time_var, id_var))
colnames(file_A) <- c("fname", "lname", "age", "id_A")
# Order rows based on fname_var, lname_var, time_var
file_A <- file_A[order(fname, lname, age), ]
# Create file_A flag
file_A$file_A <- 1
# Create ID_A column
file_A$ID_A <- 1:nrow(file_A)
# Remove duplicate rows
N <- file_A[,.N, by=c("fname", "lname", "age")]
file_A <- merge(file_A, N, by=c("fname", "lname", "age"), all.x = TRUE)
file_A <- subset(file_A, N == 1)
file_A$N <- NULL
rm(N)
# Remove rows with NA
file_A <- file_A[!is.na(fname) & !is.na(lname)]
# Return file_A
return(file_A)
}
# NOTES
# 1. Double check that the ordering given by order(...) is unique.
# 2. Add code chunk to check that id_var uniquely identifies rows.
# Define preprocess_B
#' @export
preprocess_B <- function(file_B, fname_var, lname_var, time_var, id_var) {
# Keep fname_var, lname_var, time_var, id_var
file_B <- subset(file_B, select=c(fname_var, lname_var, time_var, id_var))
colnames(file_B) <- c("fname", "lname", "age", "id_B")
# Order rows based on fname_var, lname_var, time_var
file_B <- file_B[order(fname, lname, age), ]
# Create file_B flag
file_B$file_B <- 1
# Create ID_B column
file_B$ID_B <- 1:nrow(file_B)
# Return file_B
return(file_B)
}
# NOTES
# 1. Double check that the ordering given by order(...) is unique.
# 2. Add code chunk to check that id_var uniquely identifies rows.
# Define append_A_to_B
#' @export
append_A_to_B <- function(file_A, file_B, uniqueband_file=2, backward=FALSE) {
# Append B to A
file_AB <- rbind(file_B, file_A, fill=TRUE)
# Fill in missing file flags
file_AB$file_B <- ifelse(is.na(file_AB$file_B), 0, file_AB$file_B)
file_AB$file_A <- ifelse(is.na(file_AB$file_A), 0, file_AB$file_A)
# Initialize matched_at_A, exactmatch1
file_AB$matched_at_A <- NA
file_AB$exactmatch1 <- 0
# Create count_A, count_B
if (backward) {
sum <- file_AB[,.(count_A=sum(file_B), count_B=sum(file_A)), by=c("fname", "lname", "age")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "age"), all.x = TRUE)
rm(sum)
} else {
sum <- file_AB[,.(count_A=sum(file_A), count_B=sum(file_B)), by=c("fname", "lname", "age")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "age"), all.x = TRUE)
rm(sum)
}
# Conservative ABE
if (backward) {
file_AB <- file_AB[order(file_B, fname, lname, age)]
for (i in 1:uniqueband_file) {
file_AB$temp <- 0
N <- file_AB[,.(idx=seq_len(.N), tot=.N), by=c("file_B", "fname", "lname")]
file_AB$idx <- N$idx
file_AB$tot <- N$tot
rm(N)
file_AB$age_lag <- data.table::shift(file_AB$age, n=1, fill=NA, type="lag")
file_AB$age_lead <- data.table::shift(file_AB$age, n=1, fill=NA, type="lead")
file_AB$temp <- ifelse((file_AB$age - i <= file_AB$age_lag & file_AB$idx > 1) |
(file_AB$age + i >= file_AB$age_lead & file_AB$idx < file_AB$tot), 1, file_AB$temp)
file_AB[[paste0("uniquestub_file", i)]] <- 1 - file_AB$temp
file_AB$temp <- NULL
}
file_AB$idx <- NULL
file_AB$tot <- NULL
file_AB$age_lag <- NULL
file_AB$age_lead <- NULL
} else {
file_AB <- file_AB[order(file_A, fname, lname, age)]
for (i in 1:uniqueband_file) {
file_AB$temp <- 0
N <- file_AB[,.(idx=seq_len(.N), tot=.N), by=c("file_A", "fname", "lname")]
file_AB$idx <- N$idx
file_AB$tot <- N$tot
rm(N)
file_AB$age_lag <- data.table::shift(file_AB$age, n=1, fill=NA, type="lag")
file_AB$age_lead <- data.table::shift(file_AB$age, n=1, fill=NA, type="lead")
file_AB$temp <- ifelse((file_AB$age - i <= file_AB$age_lag & file_AB$idx > 1) |
(file_AB$age + i >= file_AB$age_lead & file_AB$idx < file_AB$tot), 1, file_AB$temp)
file_AB[[paste0("uniquestub_file", i)]] <- 1 - file_AB$temp
file_AB$temp <- NULL
}
file_AB$idx <- NULL
file_AB$tot <- NULL
file_AB$age_lag <- NULL
file_AB$age_lead <- NULL
}
# Update exactmatch1
file_AB$exactmatch1 <- ifelse(file_AB$count_A == 1 & file_AB$count_B == 1, 1, file_AB$exactmatch1)
# Drop rows in A that correspond to multiple rows in B
if (backward) {
file_AB$drop <- ifelse(file_AB$count_A > 1 & file_AB$count_B == 1 & file_AB$file_A == 1, 1, 0)
} else {
file_AB$drop <- ifelse(file_AB$count_B > 1 & file_AB$count_A == 1 & file_AB$file_A == 1, 1, 0)
}
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
file_AB$count_A <- NULL
file_AB$count_B <- NULL
# Update matched_at_A
file_AB$matched_at_A <- ifelse(file_AB$exactmatch1 == 1 & file_AB$file_A == 1, 0, file_AB$matched_at_A)
# Conservative ABE
sum <- file_AB[,.(uniquestub_match0=sum(file_B)), by=c("fname", "lname", "age")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "age"), all.x = TRUE)
rm(sum)
# Return file_AB
return(file_AB)
}
# NOTES
# 1. Add a timediff parameter to be able to work with alternative time variables such as age.
# Define find_nonexact_matches
#' @export
find_nonexact_matches <- function(file_AB, timeband=2, uniqueband_match=2, backward=FALSE) {
# Check if user asked for non-exact matches
if (timeband > 0) {
# Initialize already
file_AB$already <- file_AB$exactmatch1
# Loop over timeband values
for (i in 1:timeband) {
# Create age_mi, age_pi
file_AB[[paste0("age_m", i)]] <- ifelse(file_AB$file_B == 1, file_AB$age, file_AB$age - i)
file_AB[[paste0("age_p", i)]] <- ifelse(file_AB$file_B == 1, file_AB$age, file_AB$age + i)
# Create unmatched_A
file_AB$unmatched_A <- ifelse(file_AB$file_A == 1 & file_AB$already == 0, 1, 0)
# Search -i
age_mi <- paste0("age_m", i)
sum <- file_AB[,.(mcount_A=sum(unmatched_A), mcount_B=sum(file_B), existing_matches=sum(already)), by=c("fname", "lname", age_mi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", age_mi), all.x = TRUE)
rm(sum)
# Create exactmatch1_mi
file_AB[[paste0("exactmatch1_m", i)]] <- ifelse(file_AB$mcount_A == 1 &
file_AB$mcount_B == 1 &
file_AB$existing_matches == 0, 1, NA)
# Drop rows in A-i that correspond to multiple rows in B
file_AB$drop <- ifelse(file_AB$mcount_B > 1 &
file_AB$mcount_A == 1 &
file_AB$file_A == 1 &
file_AB$existing_matches == 0, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
file_AB$existing_matches <- NULL
# Search +i
fname <- "fname"
lname <- "lname"
age_pi <- paste0("age_p", i)
sum <- file_AB[,.(pcount_A=sum(unmatched_A), pcount_B=sum(file_B), existing_matches=sum(already)), by=c("fname", "lname", age_pi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", age_pi), all.x = TRUE)
rm(sum)
# Create exactmatch1_pi
file_AB[[paste0("exactmatch1_p", i)]] <- ifelse(file_AB$pcount_A == 1 &
file_AB$pcount_B == 1 &
file_AB$existing_matches == 0, 1, NA)
# Drop rows in A+i that correspond to multiple rows in B
file_AB$drop <- ifelse(file_AB$pcount_B > 1 &
file_AB$pcount_A == 1 &
file_AB$file_A == 1 &
file_AB$existing_matches == 0, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
file_AB$existing_matches <- NULL
# Clean up exactmatch1_m, exactmatch1_p, matched_at_A
file_AB[[paste0("exactmatch1_m", i)]] <- ifelse(is.na(file_AB[[paste0("exactmatch1_m", i)]]),
0,
file_AB[[paste0("exactmatch1_m", i)]])
file_AB[[paste0("exactmatch1_p", i)]] <- ifelse(is.na(file_AB[[paste0("exactmatch1_p", i)]]),
0,
file_AB[[paste0("exactmatch1_p", i)]])
file_AB$matched_at_A <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 1 &
file_AB[[paste0("exactmatch1_m", i)]] == 0 &
file_AB$already != 1 &
file_AB$file_A == 1, i, file_AB$matched_at_A)
file_AB$matched_at_A <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 0 &
file_AB[[paste0("exactmatch1_m", i)]] == 1 &
file_AB$already != 1 &
file_AB$file_A == 1, -i, file_AB$matched_at_A)
# Update already
file_AB$already <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 1 &
file_AB[[paste0("exactmatch1_m", i)]] == 0, 1, file_AB$already)
file_AB$already <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 0 &
file_AB[[paste0("exactmatch1_m", i)]] == 1, 1, file_AB$already)
# Conservative ABE
file_AB[[paste0("uniquestub_match", i)]] <- file_AB$pcount_B + file_AB$mcount_B
file_AB[[paste0("uniquestub_match", i)]] <- ifelse(file_AB$file_A != 1,
NA,
file_AB[[paste0("uniquestub_match", i)]])
# Drop temporary columns
file_AB$pcount_B <- NULL
file_AB$pcount_A <- NULL
file_AB$mcount_B <- NULL
file_AB$mcount_A <- NULL
file_AB$unmatched_A <- NULL
}
}
# Drop unmatched in A
file_AB$drop <- ifelse(is.na(file_AB$matched_at_A) & file_AB$file_A == 1, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
# Conservative ABE
if (uniqueband_match > timeband) {
start <- timeband + 1
for (i in start:uniqueband_match) {
file_AB[[paste0("uage_m", i)]] <- file_AB$age - i
file_AB[[paste0("uage_p", i)]] <- file_AB$age + i
file_AB[[paste0("uage_m", i)]] <- ifelse(file_AB$file_B == 1,
file_AB$age,
file_AB[[paste0("uage_m", i)]])
file_AB[[paste0("uage_p", i)]] <- ifelse(file_AB$file_B == 1,
file_AB$age,
file_AB[[paste0("uage_p", i)]])
uage_mi <- paste0("uage_m", i)
sum <- file_AB[,.(unique_m=sum(file_B)), by=c("fname", "lname", uage_mi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", uage_mi), all.x = TRUE)
rm(sum)
uage_pi <- paste0("uage_p", i)
sum <- file_AB[,.(unique_m=sum(file_B)), by=c("fname", "lname", uage_pi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", uage_pi), all.x = TRUE)
rm(sum)
file_AB[[paste0("uniquestub_match", i)]] <- file_AB$unique_m + file_AB$unique_p
file_AB[[paste0("uniquestub_match", i)]] <- ifelse(file_AB$file_A != 1,
NA,
file_AB[[paste0("uniquestub_match", i)]])
file_AB[[paste0("uage_m", i)]] <- NULL
file_AB[[paste0("uage_p", i)]] <- NULL
file_AB$unique_m <- NULL
file_AB$unique_p <- NULL
}
}
# NOTES
# 1. This chunk is not tested yet.
# Conservative ABE
for (i in 1:uniqueband_match) {
j <- i - 1
file_AB[[paste0("uniquestub_match", i)]] <- file_AB[[paste0("uniquestub_match", i)]] +
file_AB[[paste0("uniquestub_match", j)]]
}
for (i in 1:uniqueband_match) {
file_AB[[paste0("uniquestub_match", i)]] <-
ifelse(file_AB$file_A == 1,
file_AB[[paste0("uniquestub_match", i)]] <= 1,
file_AB[[paste0("uniquestub_match", i)]])
}
file_AB$uniquestub_match0 <- NULL
# Drop unmatched in A
file_AB$drop <- ifelse(is.na(file_AB$matched_at_A) & file_AB$file_A == 1, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
# Create timevar_keep1
file_AB$timevar_keep1 <- ifelse(file_AB$file_A == 1, file_AB$age + file_AB$matched_at_A, file_AB$age)
# Make sure only two individuals per matched pair
sum <- file_AB[,.(count_A=sum(file_A), count_B=sum(file_B)), by=c("fname", "lname", "timevar_keep1")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "timevar_keep1"), all.x = TRUE)
rm(sum)
file_AB <- subset(file_AB, count_A == 1 & count_B == 1)
file_AB$count_A <- NULL
file_AB$count_B <- NULL
# Rename columns if backward direction (list updated for conservative ABE)
if (backward) {
colnames(file_AB) <- c("fname", "lname", "timevar_keep2", "age_p2", "age_m2", "age_p1", "age_m1",
"age", "id_A", "file_A", "ID_A", "id_B", "file_B", "ID_B", "matched_at_B",
"exactmatch2", "uniquestub_file1", "uniquestub_file2", "already",
"exactmatch2_m1", "exactmatch2_p1", "uniquestub_match1",
"exactmatch2_m2", "exactmatch2_p2", "uniquestub_match2")
}
# Return file_AB
return(file_AB)
}
# NOTES
# 1. Count number of matches at the end.
# Define fix_ids
#' @export
fix_ids <- function(file_AB_processed, timevar_keep, uniqueband_file=2, uniqueband_match=2, backward=FALSE) {
# Order rows based on fname, lname, timevar_keep, file_A
file_AB_processed <- file_AB_processed[order(file_AB_processed$fname, file_AB_processed$lname,
file_AB_processed[[timevar_keep]], file_AB_processed$file_A), ]
# Fix ID's
file_AB_processed$ID_A <- ifelse(file_AB_processed$file_B == 1,
data.table::shift(file_AB_processed$ID_A, n=1, fill=NA, type="lead"),
file_AB_processed$ID_A)
file_AB_processed$ID_B <- ifelse(file_AB_processed$file_A == 1,
data.table::shift(file_AB_processed$ID_B, n=1, fill=NA, type="lag"),
file_AB_processed$ID_B)
# Conservative ABE
if (backward) {
for (i in 1:uniqueband_match) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_match", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_B"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_B", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_match", i)]] = file_AB_processed$min
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
for (i in 1:uniqueband_file) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_file", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_B"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_B", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_file", i)]] = file_AB_processed$min
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
} else {
for (i in 1:uniqueband_match) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_match", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_A"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_A", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_match", i, "A")]] = file_AB_processed$min
file_AB_processed[[paste0("uniquestub_match", i)]] <- NULL
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
for (i in 1:uniqueband_file) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_file", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_A"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_A", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_file", i, "A")]] = file_AB_processed$min
file_AB_processed[[paste0("uniquestub_file", i)]] <- NULL
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
}
# Return file_AB_processed
return(file_AB_processed)
}
# NOTES
# 1. Add code chunk to check whether either direction has 0 rows.
#' Match records
#'
#' This function matches records in dataset A to records in dataset B
#' using the ABE method. The matched dataset includes both standard
#' and conservative ABE matches.
#'
#' @param file_A Dataset A
#' @param fname_var_A First name column in dataset A
#' @param lname_var_A Last name column in dataset A
#' @param time_var_A Time column in dataset A
#' @param id_var_A ID column in dataset A
#' @param vars_to_keep_A Other columns to be kept from dataset A
#' @param file_B Dataset B
#' @param fname_var_B First name column in dataset B
#' @param lname_var_B Last name column in dataset B
#' @param time_var_B Time column in dataset B
#' @param id_var_B ID column in dataset B
#' @param vars_to_keep_B Other columns to be kept from dataset B
#' @param out_path Path to the directory where the output file will be saved
#' @param out_file_name Output filename
#' @param timeband Time band used when searching for nonexact matches
#' @param uniqueband_file Uniqueness band used for conservative ABE (within)
#' @param uniqueband_match Uniqueness band used for conservative ABE (between)
#' @return NULL
#' @export
match_records <- function(file_A, fname_var_A, lname_var_A, time_var_A, id_var_A, vars_to_keep_A,
file_B, fname_var_B, lname_var_B, time_var_B, id_var_B, vars_to_keep_B,
out_path, out_file_name, timeband=2, uniqueband_file=2, uniqueband_match=2) {
# Forward pass
file_A_forward <- preprocess_A(file_A, fname_var_A, lname_var_A, time_var_A, id_var_A)
file_B_forward <- preprocess_B(file_B, fname_var_B, lname_var_B, time_var_B, id_var_B)
file_AB_forward <- append_A_to_B(file_A_forward, file_B_forward, uniqueband_file, FALSE)
file_AB_processed_forward <- find_nonexact_matches(file_AB_forward, timeband, uniqueband_match, FALSE)
# Backward pass
file_A_backward <- preprocess_A(file_B, fname_var_B, lname_var_B, time_var_B, id_var_B)
file_B_backward <- preprocess_B(file_A, fname_var_A, lname_var_A, time_var_A, id_var_A)
file_AB_backward <- append_A_to_B(file_A_backward, file_B_backward, uniqueband_file, TRUE)
file_AB_processed_backward <- find_nonexact_matches(file_AB_backward, timeband, uniqueband_match, TRUE)
# Fix ID's
forward <- fix_ids(file_AB_processed_forward, "timevar_keep1", uniqueband_file, uniqueband_match, FALSE)
backward <- fix_ids(file_AB_processed_backward, "timevar_keep2", uniqueband_file, uniqueband_match, TRUE)
# Keep only necessary columns (list updated for conservative ABE)
forward <- forward[, c("ID_A", "ID_B", "file_A", "matched_at_A",
"uniquestub_match1A", "uniquestub_file1A", "uniquestub_match2A", "uniquestub_file2A")]
backward <- backward[,c("ID_A", "ID_B", "file_A", "file_B", "matched_at_B",
"uniquestub_match1", "uniquestub_file1", "uniquestub_match2", "uniquestub_file2")]
# Join backward to forward
final <- merge(backward, forward, by = c("ID_A", "ID_B", "file_A"))
# NOTES
# 1. Add code chunk to check whether final has 0 rows.
# Split final to A and B and create timediff
final_A <- subset(final, file_A == 1)
final_B <- subset(final, file_B == 1)
final_A$timediff_A <- final_A$matched_at_A
final_B$timediff_B <- final_B$matched_at_B
final_A$matched_at_A <- NULL
final_A$matched_at_B <- NULL
final_B$matched_at_A <- NULL
final_B$matched_at_B <- NULL
final_A$file_A <- NULL
final_A$file_B <- NULL
final_B$file_A <- NULL
final_B$file_B <- NULL
# Merge back A
file_A <- subset(file_A, select=c(fname_var_A, lname_var_A, time_var_A, id_var_A, vars_to_keep_A))
file_A <- file_A[order(get(fname_var_A), get(lname_var_A), get(time_var_A)), ]
file_A$ID_A <- 1:nrow(file_A)
file_A <- subset(file_A, select=c("ID_A", id_var_A, fname_var_A, lname_var_A, time_var_A, vars_to_keep_A))
colnames(file_A) <- c("ID_A", "id_A", fname_var_A, lname_var_A, paste0(c(time_var_A, vars_to_keep_A), "_A"))
file_A <- merge(file_A, final_A, by = "ID_A")
# Merge back B
file_B <- subset(file_B, select=c(fname_var_B, lname_var_B, time_var_B, id_var_B, vars_to_keep_B))
file_B <- file_B[order(get(fname_var_B), get(lname_var_B), get(time_var_B)), ]
file_B$ID_B <- 1:nrow(file_B)
file_B <- subset(file_B, select=c("ID_B", id_var_B, fname_var_B, lname_var_B, time_var_B, vars_to_keep_B))
colnames(file_B) <- c("ID_B", "id_B", fname_var_B, lname_var_B, paste0(c(time_var_B, vars_to_keep_B), "_B"))
file_B <- merge(file_B, final_B, by = "ID_B")
# Merge A to B
file_A <- subset(file_A, select=c("ID_A", "id_A", fname_var_A, lname_var_A, paste0(c(time_var_A, vars_to_keep_A), "_A"), "timediff_A",
"uniquestub_match1A", "uniquestub_file1A", "uniquestub_match2A", "uniquestub_file2A",
"uniquestub_match1", "uniquestub_file1", "uniquestub_match2", "uniquestub_file2"))
file_B <- subset(file_B, select=c("ID_A", "id_B", paste0(c(time_var_B, vars_to_keep_B), "_B"), "timediff_B"))
res <- merge(file_B, file_A, by = "ID_A")
res$ID_A <- NULL
# Conservative ABE
for (i in 1:uniqueband_match) {
res[[paste0("uniquestub_match", i)]] <-
ifelse(res[[paste0("uniquestub_match", i, "A")]] < res[[paste0("uniquestub_match", i)]],
res[[paste0("uniquestub_match", i, "A")]],
res[[paste0("uniquestub_match", i)]])
res[[paste0("uniquestub_match", i, "A")]] <- NULL
}
for (i in 1:uniqueband_file) {
res[[paste0("uniquestub_file", i)]] <-
ifelse(res[[paste0("uniquestub_file", i, "A")]] < res[[paste0("uniquestub_file", i)]],
res[[paste0("uniquestub_file", i, "A")]],
res[[paste0("uniquestub_file", i)]])
res[[paste0("uniquestub_file", i, "A")]] <- NULL
}
# Save dataset
#dir.create(out_path, showWarnings = FALSE)
path_to_out_file <- paste(out_path, out_file_name, sep="/")
fwrite(res, path_to_out_file)
# Clean up after yourself
rm(list = ls())
}
caseybreen/wcensoc documentation built on Nov. 21, 2024, 5:15 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions append_A_to_B preprocess_B preprocess_A
#########################################
# created by UGUR YILDIRIM (2020-09-30) #
# revised by WON-TAK JOO (2022-11-10) #
#########################################
# Define %>%
`%>%` <- magrittr::`%>%`
# Define preprocess_A
#' @export
preprocess_A <- function(file_A, fname_var, lname_var, time_var, id_var) {
# Keep fname_var, lname_var, time_var, id_var
file_A <- subset(file_A, select=c(fname_var, lname_var, time_var, id_var))
colnames(file_A) <- c("fname", "lname", "age", "id_A")
# Order rows based on fname_var, lname_var, time_var
file_A <- file_A[order(fname, lname, age), ]
# Create file_A flag
file_A$file_A <- 1
# Create ID_A column
file_A$ID_A <- 1:nrow(file_A)
# Remove duplicate rows
N <- file_A[,.N, by=c("fname", "lname", "age")]
file_A <- merge(file_A, N, by=c("fname", "lname", "age"), all.x = TRUE)
file_A <- subset(file_A, N == 1)
file_A$N <- NULL
rm(N)
# Remove rows with NA
file_A <- file_A[!is.na(fname) & !is.na(lname)]
# Return file_A
return(file_A)
}
# NOTES
# 1. Double check that the ordering given by order(...) is unique.
# 2. Add code chunk to check that id_var uniquely identifies rows.
# Define preprocess_B
#' @export
preprocess_B <- function(file_B, fname_var, lname_var, time_var, id_var) {
# Keep fname_var, lname_var, time_var, id_var
file_B <- subset(file_B, select=c(fname_var, lname_var, time_var, id_var))
colnames(file_B) <- c("fname", "lname", "age", "id_B")
# Order rows based on fname_var, lname_var, time_var
file_B <- file_B[order(fname, lname, age), ]
# Create file_B flag
file_B$file_B <- 1
# Create ID_B column
file_B$ID_B <- 1:nrow(file_B)
# Return file_B
return(file_B)
}
# NOTES
# 1. Double check that the ordering given by order(...) is unique.
# 2. Add code chunk to check that id_var uniquely identifies rows.
# Define append_A_to_B
#' @export
append_A_to_B <- function(file_A, file_B, uniqueband_file=2, backward=FALSE) {
# Append B to A
file_AB <- rbind(file_B, file_A, fill=TRUE)
# Fill in missing file flags
file_AB$file_B <- ifelse(is.na(file_AB$file_B), 0, file_AB$file_B)
file_AB$file_A <- ifelse(is.na(file_AB$file_A), 0, file_AB$file_A)
# Initialize matched_at_A, exactmatch1
file_AB$matched_at_A <- NA
file_AB$exactmatch1 <- 0
# Create count_A, count_B
if (backward) {
sum <- file_AB[,.(count_A=sum(file_B), count_B=sum(file_A)), by=c("fname", "lname", "age")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "age"), all.x = TRUE)
rm(sum)
} else {
sum <- file_AB[,.(count_A=sum(file_A), count_B=sum(file_B)), by=c("fname", "lname", "age")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "age"), all.x = TRUE)
rm(sum)
}
# Conservative ABE
if (backward) {
file_AB <- file_AB[order(file_B, fname, lname, age)]
for (i in 1:uniqueband_file) {
file_AB$temp <- 0
N <- file_AB[,.(idx=seq_len(.N), tot=.N), by=c("file_B", "fname", "lname")]
file_AB$idx <- N$idx
file_AB$tot <- N$tot
rm(N)
file_AB$age_lag <- data.table::shift(file_AB$age, n=1, fill=NA, type="lag")
file_AB$age_lead <- data.table::shift(file_AB$age, n=1, fill=NA, type="lead")
file_AB$temp <- ifelse((file_AB$age - i <= file_AB$age_lag & file_AB$idx > 1) |
(file_AB$age + i >= file_AB$age_lead & file_AB$idx < file_AB$tot), 1, file_AB$temp)
file_AB[[paste0("uniquestub_file", i)]] <- 1 - file_AB$temp
file_AB$temp <- NULL
}
file_AB$idx <- NULL
file_AB$tot <- NULL
file_AB$age_lag <- NULL
file_AB$age_lead <- NULL
} else {
file_AB <- file_AB[order(file_A, fname, lname, age)]
for (i in 1:uniqueband_file) {
file_AB$temp <- 0
N <- file_AB[,.(idx=seq_len(.N), tot=.N), by=c("file_A", "fname", "lname")]
file_AB$idx <- N$idx
file_AB$tot <- N$tot
rm(N)
file_AB$age_lag <- data.table::shift(file_AB$age, n=1, fill=NA, type="lag")
file_AB$age_lead <- data.table::shift(file_AB$age, n=1, fill=NA, type="lead")
file_AB$temp <- ifelse((file_AB$age - i <= file_AB$age_lag & file_AB$idx > 1) |
(file_AB$age + i >= file_AB$age_lead & file_AB$idx < file_AB$tot), 1, file_AB$temp)
file_AB[[paste0("uniquestub_file", i)]] <- 1 - file_AB$temp
file_AB$temp <- NULL
}
file_AB$idx <- NULL
file_AB$tot <- NULL
file_AB$age_lag <- NULL
file_AB$age_lead <- NULL
}
# Update exactmatch1
file_AB$exactmatch1 <- ifelse(file_AB$count_A == 1 & file_AB$count_B == 1, 1, file_AB$exactmatch1)
# Drop rows in A that correspond to multiple rows in B
if (backward) {
file_AB$drop <- ifelse(file_AB$count_A > 1 & file_AB$count_B == 1 & file_AB$file_A == 1, 1, 0)
} else {
file_AB$drop <- ifelse(file_AB$count_B > 1 & file_AB$count_A == 1 & file_AB$file_A == 1, 1, 0)
}
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
file_AB$count_A <- NULL
file_AB$count_B <- NULL
# Update matched_at_A
file_AB$matched_at_A <- ifelse(file_AB$exactmatch1 == 1 & file_AB$file_A == 1, 0, file_AB$matched_at_A)
# Conservative ABE
sum <- file_AB[,.(uniquestub_match0=sum(file_B)), by=c("fname", "lname", "age")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "age"), all.x = TRUE)
rm(sum)
# Return file_AB
return(file_AB)
}
# NOTES
# 1. Add a timediff parameter to be able to work with alternative time variables such as age.
# Define find_nonexact_matches
#' @export
find_nonexact_matches <- function(file_AB, timeband=2, uniqueband_match=2, backward=FALSE) {
# Check if user asked for non-exact matches
if (timeband > 0) {
# Initialize already
file_AB$already <- file_AB$exactmatch1
# Loop over timeband values
for (i in 1:timeband) {
# Create age_mi, age_pi
file_AB[[paste0("age_m", i)]] <- ifelse(file_AB$file_B == 1, file_AB$age, file_AB$age - i)
file_AB[[paste0("age_p", i)]] <- ifelse(file_AB$file_B == 1, file_AB$age, file_AB$age + i)
# Create unmatched_A
file_AB$unmatched_A <- ifelse(file_AB$file_A == 1 & file_AB$already == 0, 1, 0)
# Search -i
age_mi <- paste0("age_m", i)
sum <- file_AB[,.(mcount_A=sum(unmatched_A), mcount_B=sum(file_B), existing_matches=sum(already)), by=c("fname", "lname", age_mi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", age_mi), all.x = TRUE)
rm(sum)
# Create exactmatch1_mi
file_AB[[paste0("exactmatch1_m", i)]] <- ifelse(file_AB$mcount_A == 1 &
file_AB$mcount_B == 1 &
file_AB$existing_matches == 0, 1, NA)
# Drop rows in A-i that correspond to multiple rows in B
file_AB$drop <- ifelse(file_AB$mcount_B > 1 &
file_AB$mcount_A == 1 &
file_AB$file_A == 1 &
file_AB$existing_matches == 0, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
file_AB$existing_matches <- NULL
# Search +i
fname <- "fname"
lname <- "lname"
age_pi <- paste0("age_p", i)
sum <- file_AB[,.(pcount_A=sum(unmatched_A), pcount_B=sum(file_B), existing_matches=sum(already)), by=c("fname", "lname", age_pi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", age_pi), all.x = TRUE)
rm(sum)
# Create exactmatch1_pi
file_AB[[paste0("exactmatch1_p", i)]] <- ifelse(file_AB$pcount_A == 1 &
file_AB$pcount_B == 1 &
file_AB$existing_matches == 0, 1, NA)
# Drop rows in A+i that correspond to multiple rows in B
file_AB$drop <- ifelse(file_AB$pcount_B > 1 &
file_AB$pcount_A == 1 &
file_AB$file_A == 1 &
file_AB$existing_matches == 0, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
file_AB$existing_matches <- NULL
# Clean up exactmatch1_m, exactmatch1_p, matched_at_A
file_AB[[paste0("exactmatch1_m", i)]] <- ifelse(is.na(file_AB[[paste0("exactmatch1_m", i)]]),
0,
file_AB[[paste0("exactmatch1_m", i)]])
file_AB[[paste0("exactmatch1_p", i)]] <- ifelse(is.na(file_AB[[paste0("exactmatch1_p", i)]]),
0,
file_AB[[paste0("exactmatch1_p", i)]])
file_AB$matched_at_A <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 1 &
file_AB[[paste0("exactmatch1_m", i)]] == 0 &
file_AB$already != 1 &
file_AB$file_A == 1, i, file_AB$matched_at_A)
file_AB$matched_at_A <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 0 &
file_AB[[paste0("exactmatch1_m", i)]] == 1 &
file_AB$already != 1 &
file_AB$file_A == 1, -i, file_AB$matched_at_A)
# Update already
file_AB$already <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 1 &
file_AB[[paste0("exactmatch1_m", i)]] == 0, 1, file_AB$already)
file_AB$already <- ifelse(file_AB[[paste0("exactmatch1_p", i)]] == 0 &
file_AB[[paste0("exactmatch1_m", i)]] == 1, 1, file_AB$already)
# Conservative ABE
file_AB[[paste0("uniquestub_match", i)]] <- file_AB$pcount_B + file_AB$mcount_B
file_AB[[paste0("uniquestub_match", i)]] <- ifelse(file_AB$file_A != 1,
NA,
file_AB[[paste0("uniquestub_match", i)]])
# Drop temporary columns
file_AB$pcount_B <- NULL
file_AB$pcount_A <- NULL
file_AB$mcount_B <- NULL
file_AB$mcount_A <- NULL
file_AB$unmatched_A <- NULL
}
}
# Drop unmatched in A
file_AB$drop <- ifelse(is.na(file_AB$matched_at_A) & file_AB$file_A == 1, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
# Conservative ABE
if (uniqueband_match > timeband) {
start <- timeband + 1
for (i in start:uniqueband_match) {
file_AB[[paste0("uage_m", i)]] <- file_AB$age - i
file_AB[[paste0("uage_p", i)]] <- file_AB$age + i
file_AB[[paste0("uage_m", i)]] <- ifelse(file_AB$file_B == 1,
file_AB$age,
file_AB[[paste0("uage_m", i)]])
file_AB[[paste0("uage_p", i)]] <- ifelse(file_AB$file_B == 1,
file_AB$age,
file_AB[[paste0("uage_p", i)]])
uage_mi <- paste0("uage_m", i)
sum <- file_AB[,.(unique_m=sum(file_B)), by=c("fname", "lname", uage_mi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", uage_mi), all.x = TRUE)
rm(sum)
uage_pi <- paste0("uage_p", i)
sum <- file_AB[,.(unique_m=sum(file_B)), by=c("fname", "lname", uage_pi)]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", uage_pi), all.x = TRUE)
rm(sum)
file_AB[[paste0("uniquestub_match", i)]] <- file_AB$unique_m + file_AB$unique_p
file_AB[[paste0("uniquestub_match", i)]] <- ifelse(file_AB$file_A != 1,
NA,
file_AB[[paste0("uniquestub_match", i)]])
file_AB[[paste0("uage_m", i)]] <- NULL
file_AB[[paste0("uage_p", i)]] <- NULL
file_AB$unique_m <- NULL
file_AB$unique_p <- NULL
}
}
# NOTES
# 1. This chunk is not tested yet.
# Conservative ABE
for (i in 1:uniqueband_match) {
j <- i - 1
file_AB[[paste0("uniquestub_match", i)]] <- file_AB[[paste0("uniquestub_match", i)]] +
file_AB[[paste0("uniquestub_match", j)]]
}
for (i in 1:uniqueband_match) {
file_AB[[paste0("uniquestub_match", i)]] <-
ifelse(file_AB$file_A == 1,
file_AB[[paste0("uniquestub_match", i)]] <= 1,
file_AB[[paste0("uniquestub_match", i)]])
}
file_AB$uniquestub_match0 <- NULL
# Drop unmatched in A
file_AB$drop <- ifelse(is.na(file_AB$matched_at_A) & file_AB$file_A == 1, 1, 0)
file_AB <- subset(file_AB, drop == 0)
file_AB$drop <- NULL
# Create timevar_keep1
file_AB$timevar_keep1 <- ifelse(file_AB$file_A == 1, file_AB$age + file_AB$matched_at_A, file_AB$age)
# Make sure only two individuals per matched pair
sum <- file_AB[,.(count_A=sum(file_A), count_B=sum(file_B)), by=c("fname", "lname", "timevar_keep1")]
file_AB <- merge(file_AB, sum, by=c("fname", "lname", "timevar_keep1"), all.x = TRUE)
rm(sum)
file_AB <- subset(file_AB, count_A == 1 & count_B == 1)
file_AB$count_A <- NULL
file_AB$count_B <- NULL
# Rename columns if backward direction (list updated for conservative ABE)
if (backward) {
colnames(file_AB) <- c("fname", "lname", "timevar_keep2", "age_p2", "age_m2", "age_p1", "age_m1",
"age", "id_A", "file_A", "ID_A", "id_B", "file_B", "ID_B", "matched_at_B",
"exactmatch2", "uniquestub_file1", "uniquestub_file2", "already",
"exactmatch2_m1", "exactmatch2_p1", "uniquestub_match1",
"exactmatch2_m2", "exactmatch2_p2", "uniquestub_match2")
}
# Return file_AB
return(file_AB)
}
# NOTES
# 1. Count number of matches at the end.
# Define fix_ids
#' @export
fix_ids <- function(file_AB_processed, timevar_keep, uniqueband_file=2, uniqueband_match=2, backward=FALSE) {
# Order rows based on fname, lname, timevar_keep, file_A
file_AB_processed <- file_AB_processed[order(file_AB_processed$fname, file_AB_processed$lname,
file_AB_processed[[timevar_keep]], file_AB_processed$file_A), ]
# Fix ID's
file_AB_processed$ID_A <- ifelse(file_AB_processed$file_B == 1,
data.table::shift(file_AB_processed$ID_A, n=1, fill=NA, type="lead"),
file_AB_processed$ID_A)
file_AB_processed$ID_B <- ifelse(file_AB_processed$file_A == 1,
data.table::shift(file_AB_processed$ID_B, n=1, fill=NA, type="lag"),
file_AB_processed$ID_B)
# Conservative ABE
if (backward) {
for (i in 1:uniqueband_match) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_match", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_B"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_B", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_match", i)]] = file_AB_processed$min
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
for (i in 1:uniqueband_file) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_file", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_B"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_B", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_file", i)]] = file_AB_processed$min
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
} else {
for (i in 1:uniqueband_match) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_match", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_A"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_A", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_match", i, "A")]] = file_AB_processed$min
file_AB_processed[[paste0("uniquestub_match", i)]] <- NULL
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
for (i in 1:uniqueband_file) {
file_AB_processed$temp <- file_AB_processed[[paste0("uniquestub_file", i)]]
min <- file_AB_processed[,.(min=min(temp, na.rm=TRUE)), by="ID_A"]
file_AB_processed <- merge(file_AB_processed, min, by="ID_A", all.x = TRUE)
rm(min)
file_AB_processed[[paste0("uniquestub_file", i, "A")]] = file_AB_processed$min
file_AB_processed[[paste0("uniquestub_file", i)]] <- NULL
file_AB_processed$temp <- NULL
file_AB_processed$min <- NULL
}
}
# Return file_AB_processed
return(file_AB_processed)
}
# NOTES
# 1. Add code chunk to check whether either direction has 0 rows.
#' Match records
#'
#' This function matches records in dataset A to records in dataset B
#' using the ABE method. The matched dataset includes both standard
#' and conservative ABE matches.
#'
#' @param file_A Dataset A
#' @param fname_var_A First name column in dataset A
#' @param lname_var_A Last name column in dataset A
#' @param time_var_A Time column in dataset A
#' @param id_var_A ID column in dataset A
#' @param vars_to_keep_A Other columns to be kept from dataset A
#' @param file_B Dataset B
#' @param fname_var_B First name column in dataset B
#' @param lname_var_B Last name column in dataset B
#' @param time_var_B Time column in dataset B
#' @param id_var_B ID column in dataset B
#' @param vars_to_keep_B Other columns to be kept from dataset B
#' @param out_path Path to the directory where the output file will be saved
#' @param out_file_name Output filename
#' @param timeband Time band used when searching for nonexact matches
#' @param uniqueband_file Uniqueness band used for conservative ABE (within)
#' @param uniqueband_match Uniqueness band used for conservative ABE (between)
#' @return NULL
#' @export
match_records <- function(file_A, fname_var_A, lname_var_A, time_var_A, id_var_A, vars_to_keep_A,
file_B, fname_var_B, lname_var_B, time_var_B, id_var_B, vars_to_keep_B,
out_path, out_file_name, timeband=2, uniqueband_file=2, uniqueband_match=2) {
# Forward pass
file_A_forward <- preprocess_A(file_A, fname_var_A, lname_var_A, time_var_A, id_var_A)
file_B_forward <- preprocess_B(file_B, fname_var_B, lname_var_B, time_var_B, id_var_B)
file_AB_forward <- append_A_to_B(file_A_forward, file_B_forward, uniqueband_file, FALSE)
file_AB_processed_forward <- find_nonexact_matches(file_AB_forward, timeband, uniqueband_match, FALSE)
# Backward pass
file_A_backward <- preprocess_A(file_B, fname_var_B, lname_var_B, time_var_B, id_var_B)
file_B_backward <- preprocess_B(file_A, fname_var_A, lname_var_A, time_var_A, id_var_A)
file_AB_backward <- append_A_to_B(file_A_backward, file_B_backward, uniqueband_file, TRUE)
file_AB_processed_backward <- find_nonexact_matches(file_AB_backward, timeband, uniqueband_match, TRUE)
# Fix ID's
forward <- fix_ids(file_AB_processed_forward, "timevar_keep1", uniqueband_file, uniqueband_match, FALSE)
backward <- fix_ids(file_AB_processed_backward, "timevar_keep2", uniqueband_file, uniqueband_match, TRUE)
# Keep only necessary columns (list updated for conservative ABE)
forward <- forward[, c("ID_A", "ID_B", "file_A", "matched_at_A",
"uniquestub_match1A", "uniquestub_file1A", "uniquestub_match2A", "uniquestub_file2A")]
backward <- backward[,c("ID_A", "ID_B", "file_A", "file_B", "matched_at_B",
"uniquestub_match1", "uniquestub_file1", "uniquestub_match2", "uniquestub_file2")]
# Join backward to forward
final <- merge(backward, forward, by = c("ID_A", "ID_B", "file_A"))
# NOTES
# 1. Add code chunk to check whether final has 0 rows.
# Split final to A and B and create timediff
final_A <- subset(final, file_A == 1)
final_B <- subset(final, file_B == 1)
final_A$timediff_A <- final_A$matched_at_A
final_B$timediff_B <- final_B$matched_at_B
final_A$matched_at_A <- NULL
final_A$matched_at_B <- NULL
final_B$matched_at_A <- NULL
final_B$matched_at_B <- NULL
final_A$file_A <- NULL
final_A$file_B <- NULL
final_B$file_A <- NULL
final_B$file_B <- NULL
# Merge back A
file_A <- subset(file_A, select=c(fname_var_A, lname_var_A, time_var_A, id_var_A, vars_to_keep_A))
file_A <- file_A[order(get(fname_var_A), get(lname_var_A), get(time_var_A)), ]
file_A$ID_A <- 1:nrow(file_A)
file_A <- subset(file_A, select=c("ID_A", id_var_A, fname_var_A, lname_var_A, time_var_A, vars_to_keep_A))
colnames(file_A) <- c("ID_A", "id_A", fname_var_A, lname_var_A, paste0(c(time_var_A, vars_to_keep_A), "_A"))
file_A <- merge(file_A, final_A, by = "ID_A")
# Merge back B
file_B <- subset(file_B, select=c(fname_var_B, lname_var_B, time_var_B, id_var_B, vars_to_keep_B))
file_B <- file_B[order(get(fname_var_B), get(lname_var_B), get(time_var_B)), ]
file_B$ID_B <- 1:nrow(file_B)
file_B <- subset(file_B, select=c("ID_B", id_var_B, fname_var_B, lname_var_B, time_var_B, vars_to_keep_B))
colnames(file_B) <- c("ID_B", "id_B", fname_var_B, lname_var_B, paste0(c(time_var_B, vars_to_keep_B), "_B"))
file_B <- merge(file_B, final_B, by = "ID_B")
# Merge A to B
file_A <- subset(file_A, select=c("ID_A", "id_A", fname_var_A, lname_var_A, paste0(c(time_var_A, vars_to_keep_A), "_A"), "timediff_A",
"uniquestub_match1A", "uniquestub_file1A", "uniquestub_match2A", "uniquestub_file2A",
"uniquestub_match1", "uniquestub_file1", "uniquestub_match2", "uniquestub_file2"))
file_B <- subset(file_B, select=c("ID_A", "id_B", paste0(c(time_var_B, vars_to_keep_B), "_B"), "timediff_B"))
res <- merge(file_B, file_A, by = "ID_A")
res$ID_A <- NULL
# Conservative ABE
for (i in 1:uniqueband_match) {
res[[paste0("uniquestub_match", i)]] <-
ifelse(res[[paste0("uniquestub_match", i, "A")]] < res[[paste0("uniquestub_match", i)]],
res[[paste0("uniquestub_match", i, "A")]],
res[[paste0("uniquestub_match", i)]])
res[[paste0("uniquestub_match", i, "A")]] <- NULL
}
for (i in 1:uniqueband_file) {
res[[paste0("uniquestub_file", i)]] <-
ifelse(res[[paste0("uniquestub_file", i, "A")]] < res[[paste0("uniquestub_file", i)]],
res[[paste0("uniquestub_file", i, "A")]],
res[[paste0("uniquestub_file", i)]])
res[[paste0("uniquestub_file", i, "A")]] <- NULL
}
# Save dataset
#dir.create(out_path, showWarnings = FALSE)
path_to_out_file <- paste(out_path, out_file_name, sep="/")
fwrite(res, path_to_out_file)
# Clean up after yourself
rm(list = ls())
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.