Nothing
R/slide.R
In DataCombine: Tools for Easily Combining and Cleaning Data Sets
#' A function for creating lag and lead variables, including for time-series
#' cross-sectional data.
#'
#' \code{slide} a function for creating lag and lead variables, including for
#' time-series cross-sectional data.
#'
#' @param data a data frame object.
#' @param Var a character string naming the variable you would like to slide
#' (create lag or lead).
#' @param TimeVar optional character string naming the time variable. If
#' specified then the data is ordered by Var-TimeVar before sliding.
#' @param GroupVar a character string naming the variable grouping the units
#' within which \code{Var} will be slid. If \code{GroupVar} is missing then the
#' whole variable is slid up or down. This is similar to \code{\link{shift}},
#' though \code{shift} returns the slid data to a new vector rather than the
#' original data frame.
#' @param NewVar a character string specifying the name for the new variable to
#' place the slid data in.
#' @param slideBy numeric value specifying how many rows (time units) to shift
#' the data by. Negative values slide the data down--lag the data. Positive
#' values shift the data up--lead the data.
#' @param keepInvalid logical. Whether or not to keep observations for groups for
#' which no valid lag/lead can be created due to an insufficient number of time
#' period observations. If \code{TRUE} then these groups are returned to the
#' bottom of the data frame and \code{NA} is given for their new lag/lead
#' variable value.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{slide}, plus other
#' messages.
#'
#' @examples
#' # Create dummy data
#' A <- B <- C <- sample(1:20, size = 20, replace = TRUE)
#' ID <- sort(rep(seq(1:4), 5))
#' Data <- data.frame(ID, A, B, C)
#'
#' # Lead the variable by two time units
#' DataSlid1 <- slide(Data, Var = 'A', NewVar = 'ALead', slideBy = 2)
#'
#' # Lag the variable one time unit by ID group
#' DataSlid2 <- slide(data = Data, Var = 'B', GroupVar = 'ID',
#' NewVar = 'BLag', slideBy = -1)
#'
#' # Lag the variable one time unit by ID group, with invalid lags
#' Data <- Data[1:16, ]
#'
#' DataSlid3 <- slide(data = Data, Var = 'B', GroupVar = 'ID',
#' NewVar = 'BLag', slideBy = -2, keepInvalid = TRUE)
#'
#' @return a data frame
#'
#' @description The function slides a column up or down to create lag or lead
#' variables. If \code{GroupVar} is specified it will slide \code{Var} for each
#' group. This is important for time-series cross-section data. The slid data is
#' placed in a new variable in the original data frame.
#' Note: your data needs to be sorted by date. The date should be ascending
#' (i.e. increasing as it moves down the rows). Also, the time difference
#' between rows should be constant, e.g. days, months, years.
#'
#' @seealso \code{\link{shift}}, \code{\link{dplyr}}
#'
#' @source Partially based on TszKin Julian's \code{shift} function:
#' \url{http://ctszkin.com/2012/03/11/generating-a-laglead-variables/}
#'
#'
#' @importFrom dplyr group_by group_by_ summarize mutate ungroup %>% bind_rows
#' @export
slide <- function(data, Var, TimeVar, GroupVar, NewVar, slideBy = -1,
keepInvalid = FALSE, reminder = TRUE) {
fake <- total <- FullData <- NULL
if (!missing(GroupVar) & "data.table" %in% class(data))
stop(paste("slide does not support data.tables with Grouped variables.\n",
"Convert to data.frame and try again."), call. = F)
if (!missing(GroupVar) & "tbl_df" %in% class(data)) {
message("Converting to plain data frame from tbl_df.")
data <- as.data.frame(data)
}
# Determine if Var exists in data
DataNames <- names(data)
TestExist <- Var %in% DataNames
if (!isTRUE(TestExist)) {
stop(paste(Var, "was not found in the data frame."),
call. = FALSE)
}
if (!missing(TimeVar))
data[order(data[, Var], data[, TimeVar]), ]
VarVect <- data[, Var]
# Variable Name
if (missing(NewVar)) {
NewVar <- paste0(Var, slideBy)
}
# Logically valid argument pairs
if (isTRUE(keepInvalid) & missing(GroupVar)) {
warning("keepInvalid set to FALSE when GroupVar is missing.")
keepInvalid <- FALSE
}
# Give messages
if (isTRUE(reminder)) {
if (missing(TimeVar)) {
if (missing(GroupVar)) {
message(paste("\nRemember to put", deparse(substitute(data)),
"in time order before running."))
}
if (!missing(GroupVar)) {
message(paste("\nRemember to order", deparse(substitute(data)),
"by", GroupVar, "and the time variable before running."))
}
}
if (slideBy < 0) {
message(paste("\nLagging", Var, "by", abs(slideBy),
"time units.\n"))
} else if (slideBy > 0) {
message(paste("\nLeading", Var, "by", abs(slideBy),
"time units.\n"))
}
}
# Drop if there are not enough observations per group to
# slide
if (!missing(GroupVar)) {
data <- group_by_(data, .dots = GroupVar)
data$fake <- 1
Minimum <- abs(slideBy) - 1
Summed <- dplyr::mutate(data, total = sum(fake))
SubSummed <- subset(Summed, total <= Minimum) %>% data.frame()
data <- VarDrop(data, "fake")
if (nrow(SubSummed) == 0) {
FullData <- NULL
} else if (nrow(SubSummed) > 0) {
## Hack
FullData <- data
class(FullData) <- "data.frame"
## End Hack
Dropping <- unique(SubSummed[, GroupVar])
## Hack
class(data) <- "data.frame"
data <- data[!(data[, GroupVar] %in% Dropping), ]
data <- group_by_(data, .dots = GroupVar)
## End Hack
if (!isTRUE(keepInvalid)) {
message(paste0("\nWarning: the following groups have ",
Minimum, " or fewer observations.", "\nNo valid lag/lead can be created, so they are dropped:\n"))
message(paste(Dropping, collapse = "\n"))
message("\n")
} else if (isTRUE(keepInvalid)) {
message(paste0("\nWarning: the following groups have ",
Minimum, " or fewer observations.", "\n No valid lag/lead can be created.",
"\n NA will be returned for these observations in the new lag/lead variable.",
"\n They will be returned at the bottom of the data frame.\n"))
message(paste(Dropping, collapse = "\n"))
message("\n")
}
}
}
# Create lags/leads
if (missing(GroupVar)) {
data[, NewVar] <- shift(VarVect = VarVect, shiftBy = slideBy,
reminder = FALSE)
} else if (!missing(GroupVar)) {
DataSub <- eval(parse(text = paste0("group_by(data[, c(GroupVar, Var)], ",
GroupVar, ")")))
vars <- eval(parse(text = paste0("dplyr::mutate(DataSub, NewVarX = shift(",
Var, ",", slideBy, ", reminder = FALSE))")))
data[, NewVar] <- vars$NewVarX
}
if (isTRUE(keepInvalid) & !is.null(FullData)) {
invalid <- FullData[(FullData[, GroupVar] %in% Dropping), ]
invalid[, NewVar] <- NA
data <- bind_rows(data, invalid)
}
data <- ungroup(data)
class(data) <- "data.frame"
return(data)
}
#' A function for creating lag and lead variables.
#'
#' \code{shift} a function for creating lag and lead variables, including for
#' time-series cross-sectional data.
#'
#' @param VarVect a vector you would like to shift (create lag or lead).
#' @param shiftBy numeric value specifying how many rows (time units) to shift
#' the data by. Negative values shift the data down--lag the data. Positive
#' values shift the data up--lead the data.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{shift}.
#'
#' @return a vector
#'
#' @description The function shifts a vector up or down to create lag or lead
#' variables.
#' Note: your data needs to be sorted by date. The date should be ascending
#' (i.e. increasing as it moves down the rows).
#'
#' @seealso \code{\link{slide}}
#'
#' @source Largely based on TszKin Julian's \code{shift} function:
#' \url{http://ctszkin.com/2012/03/11/generating-a-laglead-variables/}.
#'
#' @importFrom utils tail head
#'
#' @export
shift <- function(VarVect, shiftBy, reminder = TRUE) {
if (!is.numeric(shiftBy)) {
stop(paste(shiftBy, "must be numeric."), call. = FALSE)
}
if (isTRUE(reminder)) {
message(paste("Remember to put", deparse(substitute(data)),
"in time order before running shift."))
}
if (length(shiftBy) > 1)
return(sapply(shiftBy, shift, Var = VarVect))
out <- NULL
abs_shiftBy = abs(shiftBy)
if (shiftBy > 0) {
out <- c(tail(VarVect, -abs_shiftBy), rep(NA, abs_shiftBy))
} else if (shiftBy < 0) {
out <- c(rep(NA, abs_shiftBy), head(VarVect, -abs_shiftBy))
} else {
out <- VarVect
}
return(out)
}
#' Create a moving average for a period before or after each time point for a
#' given variable
#'
#' @param data a data frame object.
#' @param Var a character string naming the variable you would like to create
#' the lag/lead moving averages from.
#' @param GroupVar a character string naming the variable grouping the units
#' within which \code{Var} will be turned into slid moving averages. If
#' \code{GroupVar} is missing then the whole variable is slid up or down and
#' moving averages will be created. This is similar to \code{\link{shift}},
#' though \code{shift} returns the slid data to a new vector rather than the
#' original data frame.
#' @param periodBound integer. The time point for the outer bound of the time
#' period over which to create the moving averages. The default is \code{-3},
#' i.e. the moving average period begins three time points before a given time
#' point. Can also be positive for leading moving averages.
#' @param offset integer. How many time increments away from a given time point
#' to begin the moving average time period. The default is \code{1}. Effectively
#' controls how wide the moving average window is in the other direction of
#' \code{periodBound}. Note: must be positive.
#' @param NewVar a character string specifying the name for the new variable to
#' place the slid data in.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{slideMA}.
#'
#' @details \code{slideMA} is designed to give you more control over the window
#' for creating the moving average. Think of the \code{periodBound} and
#' \code{offset} arguments working together. If for example,
#' \code{periodBound = -3} and \code{offset = 1} then the variable of interest
#' will be lagged by 2 then a moving average window of three time increments
#' around the lagged variable is found.
#'
#' @return a data frame
#'
#' @examples
#' # Create dummy data
#' A <- B <- C <- sample(1:20, size = 20, replace = TRUE)
#' ID <- sort(rep(seq(1:4), 5))
#' Data <- data.frame(ID, A, B, C)
#'
#' # Lead the variable by two time units
#' DataSlidMA1 <- slideMA(Data, Var = 'A', NewVar = 'ALead_MA',
#' periodBound = 3)
#'
#' # Lag the variable one time unit by ID group
#' DataSlidMA2 <- slideMA(data = Data, Var = 'B', GroupVar = 'ID',
#' NewVar = 'BLag_MA', periodBound = -3, offset = 2)
#'
#' @seealso \code{\link{shift}}, \code{\link{slide}}, \code{\link{dplyr}}
#' @importFrom dplyr group_by mutate
#' @export
slideMA <- function(data, Var, GroupVar, periodBound = -3, offset = 1,
NewVar, reminder = TRUE) {
slideBy <- NULL
if (isTRUE(reminder)) {
if (missing(GroupVar)) {
message(paste("\nRemember to put", deparse(substitute(data)),
"in time order before running slide.\n"))
}
if (!missing(GroupVar)) {
message(paste("\nRemember to order", deparse(substitute(data)),
"by", GroupVar, "and the time variable before running slide.\n"))
}
}
# Determine if Var exists in data
DataNames <- names(data)
TestExist <- Var %in% DataNames
if (!isTRUE(TestExist)) {
stop(paste(Var, "was not found in the data frame.\n"),
call. = FALSE)
}
VarVect <- data[, Var]
if (missing(NewVar)) {
NewVar <- paste0(Var, "MA", periodBound, "_", offset)
}
Abs = abs(periodBound)
if (periodBound < 0) {
slideBy = periodBound + abs(offset)
} else if (periodBound > 0) {
slideBy = periodBound - abs(offset)
}
# Create lags/leads moving averages
if (missing(GroupVar)) {
data[, NewVar] <- shiftMA(data[, Var], shiftBy = slideBy,
Abs = Abs, reminder = FALSE)
} else if (!missing(GroupVar)) {
DataSub <- eval(parse(text = paste0("group_by(data[, c(GroupVar, Var)], ",
GroupVar, ")")))
vars <- eval(parse(text = paste0("dplyr::mutate(DataSub, NewVarX = shiftMA(",
Var, ", shiftBy =", slideBy, ", Abs = ", Abs, ", reminder = FALSE))")))
data[, NewVar] <- vars$NewVarX
data <- data.frame(data)
}
return(data)
}
#' Internal function for slideMA
#' @param x vector
#' @param shiftBy numeric
#' @param Abs numeric
#' @param reminder logical
#'
#' @keywords internals
#' @export
shiftMA <- function(x, shiftBy, Abs, reminder) {
x <- shift(x, shiftBy, reminder = reminder)
ma(x, Abs, centre = FALSE)
}
#' Spread a dummy variable (1's and 0') over a specified time period and for
#' specified groups
#'
#' @param data a data frame object.
#' @param Var a character string naming the numeric dummy variable with values
#' 0 and 1 that you would like to spread. Can be either spread as a lag or lead.
#' @param GroupVar a character string naming the variable grouping the units
#' within which \code{Var} will be spread If \code{GroupVar} is missing then the
#' whole variable is spread up or down. This is similar to \code{\link{shift}},
#' though \code{shift} slides the data and returns it to a new vector rather
#' than the original data frame.
#' @param NewVar a character string specifying the name for the new variable to
#' place the spread dummy data in.
#' @param spreadBy numeric value specifying how many rows (time units) to spread
#' the data over. Negative values spread the data down--lag the data. Positive
#' values spread the data up--lead the data.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{SpreadDummy}.
#'
#' @examples
#' # Create dummy data
#' ID <- sort(rep(seq(1:4), 5))
#' NotVar <- rep(1:5, 4)
#' Dummy <- sample(c(0, 1), size = 20, replace = TRUE)
#' Data <- data.frame(ID, NotVar, Dummy)
#'
#' # Spread
#' DataSpread1 <- SpreadDummy(data = Data, Var = 'Dummy',
#' spreadBy = 2, reminder = FALSE)
#'
#' DataSpread2 <- SpreadDummy(data = Data, Var = 'Dummy', GroupVar = 'ID',
#' spreadBy = -2)
#'
#' @seealso \code{\link{slide}}
#' @export
SpreadDummy <- function(data, Var, GroupVar, NewVar, spreadBy = -2,
reminder = TRUE) {
# Coerce to data.frame
data <- as.data.frame(data)
# Check if variable is numeric dummy
if (class(data[, Var]) != "numeric") {
stop(paste(Var, "must be a numeric dummy variable."),
call. = FALSE)
}
if (missing(NewVar)) {
NewVar <- paste0(Var, "Spread", spreadBy)
}
if (spreadBy < 0) {
start = -1
} else if (spreadBy > 0) {
start = 1
}
for (i in start:spreadBy) {
NewTemp <- paste0(NewVar, i)
if (isTRUE(reminder) & abs(i) == 1) {
if (!missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, GroupVar = GroupVar,
NewVar = NewTemp, slideBy = i))
} else if (missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, NewVar = NewTemp,
slideBy = i))
}
} else if (!missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, GroupVar = GroupVar,
NewVar = NewTemp, slideBy = i, reminder = FALSE))
} else if (missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, NewVar = NewTemp,
slideBy = i, reminder = FALSE))
}
MainNames <- names(data)
if (nrow(temp) != nrow(data)) {
data <- data[(data[, GroupVar] %in% temp[, GroupVar]),
]
}
temp <- temp[, NewTemp]
data <- data.frame(data, temp)
names(data) <- c(MainNames, NewTemp)
}
tempNames <- Var
for (i in start:spreadBy) {
temp <- paste0(NewVar, i)
tempNames <- append(tempNames, temp)
}
data[, NewVar] <- data[, Var]
for (i in tempNames) {
data[, NewVar][data[, i] == 1] <- 1
}
data <- VarDrop(data, tempNames[-1])
return(data)
}
#' Find the starting and ending time points of a spell
#'
#' \code{StartEnd} finds the starting and ending time points of a spell,
#' including for time-series cross-sectional data. Note: your data needs to be
#' sorted by date. The date should be ascending (i.e. increasing as it moves
#' down the rows).
#'
#' @param data a data frame object.
#' @param SpellVar a character string naming the variable with information on
#' when each spell starts.
#' @param GroupVar a character string naming the variable grouping the units
#' experiencing the spells. If \code{GroupVar} is missing then .
#' @param SpellValue a value indicating when a unit is in a spell. If
#' \code{SpellValue} is missing then any change in \code{Var}'s value will be
#' treated as the start/end of a spell. Must specify if \code{OnlyStart = TRUE}.
#' @param OnlyStart logical for whether or not to only add a new
#' \code{Spell_Start} variable. Please see the details.
#' @param ... Aguments to pass to \code{\link{slide}}.
#'
#' @examples
#' # Create fake data
#' ID <- sort(rep(seq(1:4), 5))
#' Time <- rep(1:5, 4)
#' Dummy <- c(1, sample(c(0, 1), size = 19, replace = TRUE))
#' Data <- data.frame(ID, Time, Dummy)
#'
#' # Find start/end of spells denoted by Dummy = 1
#' DataSpell <- StartEnd(Data, SpellVar = 'Dummy', GroupVar = 'ID',
#' TimeVar = 'Time', SpellValue = 1)
#'
#' head(DataSpell)
#'
#' @return a data frame. If \code{OnlyStart = FALSE} then two new variables are
#' returned:
#' \itemize{
#' \item{Spell_Start: }{The time period year of a given spell.}
#' \item{Spell_End: }{The end time period of a given spell.}
#' }
#' If \code{OnlyStart = TRUE} then only \code{Spell_Start} is added.
#' This variable includes both \code{1}'s for the start of a new spell and for
#' the start of a 'gap spell', i.e. a spell after \code{Spell_End}.
#' @seealso \code{\link{slide}}
#' @importFrom dplyr ungroup
#' @export
StartEnd <- function(data, SpellVar, GroupVar, SpellValue, OnlyStart = FALSE,
...) {
if (missing(SpellVar)) {
stop("You must specify the SpellVar", call. = FALSE)
}
if (!missing(GroupVar)) {
message(paste("\nRemember to order", deparse(substitute(data)),
"by", GroupVar, "and the time variable before running slide.\n"))
}
# Find Start
Temp <- slide(data = data, Var = SpellVar, GroupVar = GroupVar,
slideBy = -1, NewVar = "TempStart", reminder = FALSE,
...)
Temp <- data.frame(Temp)
Temp$Spell_Start <- 0
Temp$Spell_Start[is.na(Temp[, SpellVar])] <- NA
if (missing(SpellValue)) {
Temp$Spell_Start[Temp[, SpellVar] != Temp[, "TempStart"]] <- 1
} else if (!missing(SpellValue)) {
Temp$Spell_Start[Temp[, SpellVar] == SpellValue & Temp[,
"TempStart"] != SpellValue] <- 1
}
# Find End
Temp <- slide(data = Temp, Var = SpellVar, GroupVar = GroupVar,
slideBy = 1, NewVar = "TempEnd", reminder = FALSE, ...)
Temp <- data.frame(Temp)
Temp$Spell_End <- 0
Temp$Spell_End[is.na(Temp[, SpellVar])] <- NA
if (missing(SpellValue)) {
Temp$Spell_End[Temp[, SpellVar] != Temp[, "TempEnd"]] <- 1
} else if (!missing(SpellValue)) {
Temp$Spell_End[Temp[, SpellVar] == SpellValue & Temp[,
"TempEnd"] != SpellValue] <- 1
}
Temp <- ungroup(Temp)
if (isTRUE(OnlyStart)) {
if (missing(SpellValue)) {
stop("Must specify SpellValue if OnlyStart = TRUE.",
call. = FALSE)
}
Temp <- slide(data = Temp, Var = "Spell_End", GroupVar = GroupVar,
slideBy = -1, NewVar = "TempNewStart", reminder = FALSE,
...)
Temp <- ungroup(Temp)
Temp$Spell_Start[Temp$Spell_Start != SpellValue & Temp$TempNewStart ==
SpellValue] <- 1
Temp <- VarDrop(Temp, c("TempStart", "TempEnd", "TempNewStart",
"Spell_End"))
} else if (!isTRUE(OnlyStart)) {
Temp <- VarDrop(Temp, c("TempStart", "TempEnd"))
}
return(Temp)
}
#' Count spells, including for grouped data
#'
#' \code{CountSpell} is a function that returns a variable counting the spell
#' number for an observation. Works with grouped data.
#' @param data a data frame object.
#' @param TimeVar a character string naming the time variable.
#' @param SpellVar a character string naming the variable with information on
#' when each spell starts.
#' @param GroupVar a character string naming the variable grouping the units
#' experiencing the spells.
#' @param NewVar NewVar a character string naming the new variable to place the
#' spell counts in.
#' @param SpellValue a value indicating when a unit is in a spell. Must match
#' the class of the \code{SpellVar}.
#'
#' @examples
#' # Create fake data
#' ID <- sort(rep(seq(1:4), 5))
#' Time <- rep(1:20)
#' Dummy <- c(1, sample(c(0, 1), size = 19, replace = TRUE))
#' Data <- data.frame(ID, Time, Dummy)
#'
#' # Find spell for whole data frame
#' DataSpell1 <- CountSpell(Data, TimeVar = 'Time', SpellVar = 'Dummy',
#' SpellValue = 1)
#'
#' head(DataSpell1)
#'
#' # Find spell for each ID group
#' DataSpell2 <- CountSpell(Data, TimeVar = 'Time', SpellVar = 'Dummy',
#' GroupVar = 'ID', SpellValue = 1)
#'
#' head(DataSpell2)
#'
#' @export
CountSpell <- function(data, TimeVar, SpellVar, GroupVar, NewVar,
SpellValue) {
if (missing(NewVar)) {
NewVar <- paste0(SpellVar, "_", "SpellCount")
message(paste0("\nSpell count placed in new variable: ",
NewVar, ".\n"))
}
if (!missing(GroupVar)) {
tempMain <- data.frame()
for (i in unique(data[, GroupVar])) {
tempSub <- data[data[, GroupVar] == i, ]
tempSub[, NewVar] <- CountSpellOne(data = tempSub,
TimeVar = TimeVar, SpellVar = SpellVar, SpellValue = SpellValue)
tempMain <- rbind(tempMain, tempSub)
}
data <- tempMain
} else if (missing(GroupVar)) {
data[, NewVar] <- CountSpellOne(data = data, TimeVar = TimeVar,
SpellVar = SpellVar, SpellValue = SpellValue)
}
return(data)
}
#' Internal function for finding spells for one unit
#'
#' @importFrom dplyr arrange_
#' @noRd
CountSpellOne <- function(data, TimeVar, SpellVar, SpellValue) {
Spell_Start <- NULL
if (missing(TimeVar))
stop("You must specify the TimeVar", call. = FALSE)
if (missing(SpellValue))
stop("You must specify the SpellValue", call. = FALSE)
if (class(SpellValue) != class(data[, SpellVar]))
stop("SpellValue must be the same class as SpellVar",
call. = FALSE)
data <- arrange_(data, .dots = TimeVar)
if (all(data[, SpellVar] == SpellValue)) {
dataSpell <- data
dataSpell$Spell_Count <- 1:nrow(dataSpell)
} else if (!any(data[, SpellVar] == SpellValue)) {
dataSpell <- data
dataSpell$Spell_Count <- 0
} else if (any(data[, SpellVar] != SpellValue)) {
dataSpell <- StartEnd(data = data, SpellVar = SpellVar,
SpellValue = SpellValue, OnlyStart = TRUE)
temp <- subset(dataSpell, Spell_Start == 1)
temp$spell_ID <- 1:nrow(temp)
dataSpell$Spell_Count <- 0
for (u in 1:max(temp$spell_ID)) {
dataSpell$Spell_Count[dataSpell[, TimeVar] >= temp[temp$spell_ID ==
u, TimeVar]] <- u
}
}
return(dataSpell$Spell_Count)
}
Try the DataCombine package in your browser
Any scripts or data that you put into this service are public.
DataCombine documentation built on May 2, 2019, 11:26 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.
#' A function for creating lag and lead variables, including for time-series
#' cross-sectional data.
#'
#' \code{slide} a function for creating lag and lead variables, including for
#' time-series cross-sectional data.
#'
#' @param data a data frame object.
#' @param Var a character string naming the variable you would like to slide
#' (create lag or lead).
#' @param TimeVar optional character string naming the time variable. If
#' specified then the data is ordered by Var-TimeVar before sliding.
#' @param GroupVar a character string naming the variable grouping the units
#' within which \code{Var} will be slid. If \code{GroupVar} is missing then the
#' whole variable is slid up or down. This is similar to \code{\link{shift}},
#' though \code{shift} returns the slid data to a new vector rather than the
#' original data frame.
#' @param NewVar a character string specifying the name for the new variable to
#' place the slid data in.
#' @param slideBy numeric value specifying how many rows (time units) to shift
#' the data by. Negative values slide the data down--lag the data. Positive
#' values shift the data up--lead the data.
#' @param keepInvalid logical. Whether or not to keep observations for groups for
#' which no valid lag/lead can be created due to an insufficient number of time
#' period observations. If \code{TRUE} then these groups are returned to the
#' bottom of the data frame and \code{NA} is given for their new lag/lead
#' variable value.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{slide}, plus other
#' messages.
#'
#' @examples
#' # Create dummy data
#' A <- B <- C <- sample(1:20, size = 20, replace = TRUE)
#' ID <- sort(rep(seq(1:4), 5))
#' Data <- data.frame(ID, A, B, C)
#'
#' # Lead the variable by two time units
#' DataSlid1 <- slide(Data, Var = 'A', NewVar = 'ALead', slideBy = 2)
#'
#' # Lag the variable one time unit by ID group
#' DataSlid2 <- slide(data = Data, Var = 'B', GroupVar = 'ID',
#' NewVar = 'BLag', slideBy = -1)
#'
#' # Lag the variable one time unit by ID group, with invalid lags
#' Data <- Data[1:16, ]
#'
#' DataSlid3 <- slide(data = Data, Var = 'B', GroupVar = 'ID',
#' NewVar = 'BLag', slideBy = -2, keepInvalid = TRUE)
#'
#' @return a data frame
#'
#' @description The function slides a column up or down to create lag or lead
#' variables. If \code{GroupVar} is specified it will slide \code{Var} for each
#' group. This is important for time-series cross-section data. The slid data is
#' placed in a new variable in the original data frame.
#' Note: your data needs to be sorted by date. The date should be ascending
#' (i.e. increasing as it moves down the rows). Also, the time difference
#' between rows should be constant, e.g. days, months, years.
#'
#' @seealso \code{\link{shift}}, \code{\link{dplyr}}
#'
#' @source Partially based on TszKin Julian's \code{shift} function:
#' \url{http://ctszkin.com/2012/03/11/generating-a-laglead-variables/}
#'
#'
#' @importFrom dplyr group_by group_by_ summarize mutate ungroup %>% bind_rows
#' @export
slide <- function(data, Var, TimeVar, GroupVar, NewVar, slideBy = -1,
keepInvalid = FALSE, reminder = TRUE) {
fake <- total <- FullData <- NULL
if (!missing(GroupVar) & "data.table" %in% class(data))
stop(paste("slide does not support data.tables with Grouped variables.\n",
"Convert to data.frame and try again."), call. = F)
if (!missing(GroupVar) & "tbl_df" %in% class(data)) {
message("Converting to plain data frame from tbl_df.")
data <- as.data.frame(data)
}
# Determine if Var exists in data
DataNames <- names(data)
TestExist <- Var %in% DataNames
if (!isTRUE(TestExist)) {
stop(paste(Var, "was not found in the data frame."),
call. = FALSE)
}
if (!missing(TimeVar))
data[order(data[, Var], data[, TimeVar]), ]
VarVect <- data[, Var]
# Variable Name
if (missing(NewVar)) {
NewVar <- paste0(Var, slideBy)
}
# Logically valid argument pairs
if (isTRUE(keepInvalid) & missing(GroupVar)) {
warning("keepInvalid set to FALSE when GroupVar is missing.")
keepInvalid <- FALSE
}
# Give messages
if (isTRUE(reminder)) {
if (missing(TimeVar)) {
if (missing(GroupVar)) {
message(paste("\nRemember to put", deparse(substitute(data)),
"in time order before running."))
}
if (!missing(GroupVar)) {
message(paste("\nRemember to order", deparse(substitute(data)),
"by", GroupVar, "and the time variable before running."))
}
}
if (slideBy < 0) {
message(paste("\nLagging", Var, "by", abs(slideBy),
"time units.\n"))
} else if (slideBy > 0) {
message(paste("\nLeading", Var, "by", abs(slideBy),
"time units.\n"))
}
}
# Drop if there are not enough observations per group to
# slide
if (!missing(GroupVar)) {
data <- group_by_(data, .dots = GroupVar)
data$fake <- 1
Minimum <- abs(slideBy) - 1
Summed <- dplyr::mutate(data, total = sum(fake))
SubSummed <- subset(Summed, total <= Minimum) %>% data.frame()
data <- VarDrop(data, "fake")
if (nrow(SubSummed) == 0) {
FullData <- NULL
} else if (nrow(SubSummed) > 0) {
## Hack
FullData <- data
class(FullData) <- "data.frame"
## End Hack
Dropping <- unique(SubSummed[, GroupVar])
## Hack
class(data) <- "data.frame"
data <- data[!(data[, GroupVar] %in% Dropping), ]
data <- group_by_(data, .dots = GroupVar)
## End Hack
if (!isTRUE(keepInvalid)) {
message(paste0("\nWarning: the following groups have ",
Minimum, " or fewer observations.", "\nNo valid lag/lead can be created, so they are dropped:\n"))
message(paste(Dropping, collapse = "\n"))
message("\n")
} else if (isTRUE(keepInvalid)) {
message(paste0("\nWarning: the following groups have ",
Minimum, " or fewer observations.", "\n No valid lag/lead can be created.",
"\n NA will be returned for these observations in the new lag/lead variable.",
"\n They will be returned at the bottom of the data frame.\n"))
message(paste(Dropping, collapse = "\n"))
message("\n")
}
}
}
# Create lags/leads
if (missing(GroupVar)) {
data[, NewVar] <- shift(VarVect = VarVect, shiftBy = slideBy,
reminder = FALSE)
} else if (!missing(GroupVar)) {
DataSub <- eval(parse(text = paste0("group_by(data[, c(GroupVar, Var)], ",
GroupVar, ")")))
vars <- eval(parse(text = paste0("dplyr::mutate(DataSub, NewVarX = shift(",
Var, ",", slideBy, ", reminder = FALSE))")))
data[, NewVar] <- vars$NewVarX
}
if (isTRUE(keepInvalid) & !is.null(FullData)) {
invalid <- FullData[(FullData[, GroupVar] %in% Dropping), ]
invalid[, NewVar] <- NA
data <- bind_rows(data, invalid)
}
data <- ungroup(data)
class(data) <- "data.frame"
return(data)
}
#' A function for creating lag and lead variables.
#'
#' \code{shift} a function for creating lag and lead variables, including for
#' time-series cross-sectional data.
#'
#' @param VarVect a vector you would like to shift (create lag or lead).
#' @param shiftBy numeric value specifying how many rows (time units) to shift
#' the data by. Negative values shift the data down--lag the data. Positive
#' values shift the data up--lead the data.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{shift}.
#'
#' @return a vector
#'
#' @description The function shifts a vector up or down to create lag or lead
#' variables.
#' Note: your data needs to be sorted by date. The date should be ascending
#' (i.e. increasing as it moves down the rows).
#'
#' @seealso \code{\link{slide}}
#'
#' @source Largely based on TszKin Julian's \code{shift} function:
#' \url{http://ctszkin.com/2012/03/11/generating-a-laglead-variables/}.
#'
#' @importFrom utils tail head
#'
#' @export
shift <- function(VarVect, shiftBy, reminder = TRUE) {
if (!is.numeric(shiftBy)) {
stop(paste(shiftBy, "must be numeric."), call. = FALSE)
}
if (isTRUE(reminder)) {
message(paste("Remember to put", deparse(substitute(data)),
"in time order before running shift."))
}
if (length(shiftBy) > 1)
return(sapply(shiftBy, shift, Var = VarVect))
out <- NULL
abs_shiftBy = abs(shiftBy)
if (shiftBy > 0) {
out <- c(tail(VarVect, -abs_shiftBy), rep(NA, abs_shiftBy))
} else if (shiftBy < 0) {
out <- c(rep(NA, abs_shiftBy), head(VarVect, -abs_shiftBy))
} else {
out <- VarVect
}
return(out)
}
#' Create a moving average for a period before or after each time point for a
#' given variable
#'
#' @param data a data frame object.
#' @param Var a character string naming the variable you would like to create
#' the lag/lead moving averages from.
#' @param GroupVar a character string naming the variable grouping the units
#' within which \code{Var} will be turned into slid moving averages. If
#' \code{GroupVar} is missing then the whole variable is slid up or down and
#' moving averages will be created. This is similar to \code{\link{shift}},
#' though \code{shift} returns the slid data to a new vector rather than the
#' original data frame.
#' @param periodBound integer. The time point for the outer bound of the time
#' period over which to create the moving averages. The default is \code{-3},
#' i.e. the moving average period begins three time points before a given time
#' point. Can also be positive for leading moving averages.
#' @param offset integer. How many time increments away from a given time point
#' to begin the moving average time period. The default is \code{1}. Effectively
#' controls how wide the moving average window is in the other direction of
#' \code{periodBound}. Note: must be positive.
#' @param NewVar a character string specifying the name for the new variable to
#' place the slid data in.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{slideMA}.
#'
#' @details \code{slideMA} is designed to give you more control over the window
#' for creating the moving average. Think of the \code{periodBound} and
#' \code{offset} arguments working together. If for example,
#' \code{periodBound = -3} and \code{offset = 1} then the variable of interest
#' will be lagged by 2 then a moving average window of three time increments
#' around the lagged variable is found.
#'
#' @return a data frame
#'
#' @examples
#' # Create dummy data
#' A <- B <- C <- sample(1:20, size = 20, replace = TRUE)
#' ID <- sort(rep(seq(1:4), 5))
#' Data <- data.frame(ID, A, B, C)
#'
#' # Lead the variable by two time units
#' DataSlidMA1 <- slideMA(Data, Var = 'A', NewVar = 'ALead_MA',
#' periodBound = 3)
#'
#' # Lag the variable one time unit by ID group
#' DataSlidMA2 <- slideMA(data = Data, Var = 'B', GroupVar = 'ID',
#' NewVar = 'BLag_MA', periodBound = -3, offset = 2)
#'
#' @seealso \code{\link{shift}}, \code{\link{slide}}, \code{\link{dplyr}}
#' @importFrom dplyr group_by mutate
#' @export
slideMA <- function(data, Var, GroupVar, periodBound = -3, offset = 1,
NewVar, reminder = TRUE) {
slideBy <- NULL
if (isTRUE(reminder)) {
if (missing(GroupVar)) {
message(paste("\nRemember to put", deparse(substitute(data)),
"in time order before running slide.\n"))
}
if (!missing(GroupVar)) {
message(paste("\nRemember to order", deparse(substitute(data)),
"by", GroupVar, "and the time variable before running slide.\n"))
}
}
# Determine if Var exists in data
DataNames <- names(data)
TestExist <- Var %in% DataNames
if (!isTRUE(TestExist)) {
stop(paste(Var, "was not found in the data frame.\n"),
call. = FALSE)
}
VarVect <- data[, Var]
if (missing(NewVar)) {
NewVar <- paste0(Var, "MA", periodBound, "_", offset)
}
Abs = abs(periodBound)
if (periodBound < 0) {
slideBy = periodBound + abs(offset)
} else if (periodBound > 0) {
slideBy = periodBound - abs(offset)
}
# Create lags/leads moving averages
if (missing(GroupVar)) {
data[, NewVar] <- shiftMA(data[, Var], shiftBy = slideBy,
Abs = Abs, reminder = FALSE)
} else if (!missing(GroupVar)) {
DataSub <- eval(parse(text = paste0("group_by(data[, c(GroupVar, Var)], ",
GroupVar, ")")))
vars <- eval(parse(text = paste0("dplyr::mutate(DataSub, NewVarX = shiftMA(",
Var, ", shiftBy =", slideBy, ", Abs = ", Abs, ", reminder = FALSE))")))
data[, NewVar] <- vars$NewVarX
data <- data.frame(data)
}
return(data)
}
#' Internal function for slideMA
#' @param x vector
#' @param shiftBy numeric
#' @param Abs numeric
#' @param reminder logical
#'
#' @keywords internals
#' @export
shiftMA <- function(x, shiftBy, Abs, reminder) {
x <- shift(x, shiftBy, reminder = reminder)
ma(x, Abs, centre = FALSE)
}
#' Spread a dummy variable (1's and 0') over a specified time period and for
#' specified groups
#'
#' @param data a data frame object.
#' @param Var a character string naming the numeric dummy variable with values
#' 0 and 1 that you would like to spread. Can be either spread as a lag or lead.
#' @param GroupVar a character string naming the variable grouping the units
#' within which \code{Var} will be spread If \code{GroupVar} is missing then the
#' whole variable is spread up or down. This is similar to \code{\link{shift}},
#' though \code{shift} slides the data and returns it to a new vector rather
#' than the original data frame.
#' @param NewVar a character string specifying the name for the new variable to
#' place the spread dummy data in.
#' @param spreadBy numeric value specifying how many rows (time units) to spread
#' the data over. Negative values spread the data down--lag the data. Positive
#' values spread the data up--lead the data.
#' @param reminder logical. Whether or not to remind you to order your data by
#' the \code{GroupVar} and time variable before running \code{SpreadDummy}.
#'
#' @examples
#' # Create dummy data
#' ID <- sort(rep(seq(1:4), 5))
#' NotVar <- rep(1:5, 4)
#' Dummy <- sample(c(0, 1), size = 20, replace = TRUE)
#' Data <- data.frame(ID, NotVar, Dummy)
#'
#' # Spread
#' DataSpread1 <- SpreadDummy(data = Data, Var = 'Dummy',
#' spreadBy = 2, reminder = FALSE)
#'
#' DataSpread2 <- SpreadDummy(data = Data, Var = 'Dummy', GroupVar = 'ID',
#' spreadBy = -2)
#'
#' @seealso \code{\link{slide}}
#' @export
SpreadDummy <- function(data, Var, GroupVar, NewVar, spreadBy = -2,
reminder = TRUE) {
# Coerce to data.frame
data <- as.data.frame(data)
# Check if variable is numeric dummy
if (class(data[, Var]) != "numeric") {
stop(paste(Var, "must be a numeric dummy variable."),
call. = FALSE)
}
if (missing(NewVar)) {
NewVar <- paste0(Var, "Spread", spreadBy)
}
if (spreadBy < 0) {
start = -1
} else if (spreadBy > 0) {
start = 1
}
for (i in start:spreadBy) {
NewTemp <- paste0(NewVar, i)
if (isTRUE(reminder) & abs(i) == 1) {
if (!missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, GroupVar = GroupVar,
NewVar = NewTemp, slideBy = i))
} else if (missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, NewVar = NewTemp,
slideBy = i))
}
} else if (!missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, GroupVar = GroupVar,
NewVar = NewTemp, slideBy = i, reminder = FALSE))
} else if (missing(GroupVar)) {
temp <- suppressMessages(slide(data, Var = Var, NewVar = NewTemp,
slideBy = i, reminder = FALSE))
}
MainNames <- names(data)
if (nrow(temp) != nrow(data)) {
data <- data[(data[, GroupVar] %in% temp[, GroupVar]),
]
}
temp <- temp[, NewTemp]
data <- data.frame(data, temp)
names(data) <- c(MainNames, NewTemp)
}
tempNames <- Var
for (i in start:spreadBy) {
temp <- paste0(NewVar, i)
tempNames <- append(tempNames, temp)
}
data[, NewVar] <- data[, Var]
for (i in tempNames) {
data[, NewVar][data[, i] == 1] <- 1
}
data <- VarDrop(data, tempNames[-1])
return(data)
}
#' Find the starting and ending time points of a spell
#'
#' \code{StartEnd} finds the starting and ending time points of a spell,
#' including for time-series cross-sectional data. Note: your data needs to be
#' sorted by date. The date should be ascending (i.e. increasing as it moves
#' down the rows).
#'
#' @param data a data frame object.
#' @param SpellVar a character string naming the variable with information on
#' when each spell starts.
#' @param GroupVar a character string naming the variable grouping the units
#' experiencing the spells. If \code{GroupVar} is missing then .
#' @param SpellValue a value indicating when a unit is in a spell. If
#' \code{SpellValue} is missing then any change in \code{Var}'s value will be
#' treated as the start/end of a spell. Must specify if \code{OnlyStart = TRUE}.
#' @param OnlyStart logical for whether or not to only add a new
#' \code{Spell_Start} variable. Please see the details.
#' @param ... Aguments to pass to \code{\link{slide}}.
#'
#' @examples
#' # Create fake data
#' ID <- sort(rep(seq(1:4), 5))
#' Time <- rep(1:5, 4)
#' Dummy <- c(1, sample(c(0, 1), size = 19, replace = TRUE))
#' Data <- data.frame(ID, Time, Dummy)
#'
#' # Find start/end of spells denoted by Dummy = 1
#' DataSpell <- StartEnd(Data, SpellVar = 'Dummy', GroupVar = 'ID',
#' TimeVar = 'Time', SpellValue = 1)
#'
#' head(DataSpell)
#'
#' @return a data frame. If \code{OnlyStart = FALSE} then two new variables are
#' returned:
#' \itemize{
#' \item{Spell_Start: }{The time period year of a given spell.}
#' \item{Spell_End: }{The end time period of a given spell.}
#' }
#' If \code{OnlyStart = TRUE} then only \code{Spell_Start} is added.
#' This variable includes both \code{1}'s for the start of a new spell and for
#' the start of a 'gap spell', i.e. a spell after \code{Spell_End}.
#' @seealso \code{\link{slide}}
#' @importFrom dplyr ungroup
#' @export
StartEnd <- function(data, SpellVar, GroupVar, SpellValue, OnlyStart = FALSE,
...) {
if (missing(SpellVar)) {
stop("You must specify the SpellVar", call. = FALSE)
}
if (!missing(GroupVar)) {
message(paste("\nRemember to order", deparse(substitute(data)),
"by", GroupVar, "and the time variable before running slide.\n"))
}
# Find Start
Temp <- slide(data = data, Var = SpellVar, GroupVar = GroupVar,
slideBy = -1, NewVar = "TempStart", reminder = FALSE,
...)
Temp <- data.frame(Temp)
Temp$Spell_Start <- 0
Temp$Spell_Start[is.na(Temp[, SpellVar])] <- NA
if (missing(SpellValue)) {
Temp$Spell_Start[Temp[, SpellVar] != Temp[, "TempStart"]] <- 1
} else if (!missing(SpellValue)) {
Temp$Spell_Start[Temp[, SpellVar] == SpellValue & Temp[,
"TempStart"] != SpellValue] <- 1
}
# Find End
Temp <- slide(data = Temp, Var = SpellVar, GroupVar = GroupVar,
slideBy = 1, NewVar = "TempEnd", reminder = FALSE, ...)
Temp <- data.frame(Temp)
Temp$Spell_End <- 0
Temp$Spell_End[is.na(Temp[, SpellVar])] <- NA
if (missing(SpellValue)) {
Temp$Spell_End[Temp[, SpellVar] != Temp[, "TempEnd"]] <- 1
} else if (!missing(SpellValue)) {
Temp$Spell_End[Temp[, SpellVar] == SpellValue & Temp[,
"TempEnd"] != SpellValue] <- 1
}
Temp <- ungroup(Temp)
if (isTRUE(OnlyStart)) {
if (missing(SpellValue)) {
stop("Must specify SpellValue if OnlyStart = TRUE.",
call. = FALSE)
}
Temp <- slide(data = Temp, Var = "Spell_End", GroupVar = GroupVar,
slideBy = -1, NewVar = "TempNewStart", reminder = FALSE,
...)
Temp <- ungroup(Temp)
Temp$Spell_Start[Temp$Spell_Start != SpellValue & Temp$TempNewStart ==
SpellValue] <- 1
Temp <- VarDrop(Temp, c("TempStart", "TempEnd", "TempNewStart",
"Spell_End"))
} else if (!isTRUE(OnlyStart)) {
Temp <- VarDrop(Temp, c("TempStart", "TempEnd"))
}
return(Temp)
}
#' Count spells, including for grouped data
#'
#' \code{CountSpell} is a function that returns a variable counting the spell
#' number for an observation. Works with grouped data.
#' @param data a data frame object.
#' @param TimeVar a character string naming the time variable.
#' @param SpellVar a character string naming the variable with information on
#' when each spell starts.
#' @param GroupVar a character string naming the variable grouping the units
#' experiencing the spells.
#' @param NewVar NewVar a character string naming the new variable to place the
#' spell counts in.
#' @param SpellValue a value indicating when a unit is in a spell. Must match
#' the class of the \code{SpellVar}.
#'
#' @examples
#' # Create fake data
#' ID <- sort(rep(seq(1:4), 5))
#' Time <- rep(1:20)
#' Dummy <- c(1, sample(c(0, 1), size = 19, replace = TRUE))
#' Data <- data.frame(ID, Time, Dummy)
#'
#' # Find spell for whole data frame
#' DataSpell1 <- CountSpell(Data, TimeVar = 'Time', SpellVar = 'Dummy',
#' SpellValue = 1)
#'
#' head(DataSpell1)
#'
#' # Find spell for each ID group
#' DataSpell2 <- CountSpell(Data, TimeVar = 'Time', SpellVar = 'Dummy',
#' GroupVar = 'ID', SpellValue = 1)
#'
#' head(DataSpell2)
#'
#' @export
CountSpell <- function(data, TimeVar, SpellVar, GroupVar, NewVar,
SpellValue) {
if (missing(NewVar)) {
NewVar <- paste0(SpellVar, "_", "SpellCount")
message(paste0("\nSpell count placed in new variable: ",
NewVar, ".\n"))
}
if (!missing(GroupVar)) {
tempMain <- data.frame()
for (i in unique(data[, GroupVar])) {
tempSub <- data[data[, GroupVar] == i, ]
tempSub[, NewVar] <- CountSpellOne(data = tempSub,
TimeVar = TimeVar, SpellVar = SpellVar, SpellValue = SpellValue)
tempMain <- rbind(tempMain, tempSub)
}
data <- tempMain
} else if (missing(GroupVar)) {
data[, NewVar] <- CountSpellOne(data = data, TimeVar = TimeVar,
SpellVar = SpellVar, SpellValue = SpellValue)
}
return(data)
}
#' Internal function for finding spells for one unit
#'
#' @importFrom dplyr arrange_
#' @noRd
CountSpellOne <- function(data, TimeVar, SpellVar, SpellValue) {
Spell_Start <- NULL
if (missing(TimeVar))
stop("You must specify the TimeVar", call. = FALSE)
if (missing(SpellValue))
stop("You must specify the SpellValue", call. = FALSE)
if (class(SpellValue) != class(data[, SpellVar]))
stop("SpellValue must be the same class as SpellVar",
call. = FALSE)
data <- arrange_(data, .dots = TimeVar)
if (all(data[, SpellVar] == SpellValue)) {
dataSpell <- data
dataSpell$Spell_Count <- 1:nrow(dataSpell)
} else if (!any(data[, SpellVar] == SpellValue)) {
dataSpell <- data
dataSpell$Spell_Count <- 0
} else if (any(data[, SpellVar] != SpellValue)) {
dataSpell <- StartEnd(data = data, SpellVar = SpellVar,
SpellValue = SpellValue, OnlyStart = TRUE)
temp <- subset(dataSpell, Spell_Start == 1)
temp$spell_ID <- 1:nrow(temp)
dataSpell$Spell_Count <- 0
for (u in 1:max(temp$spell_ID)) {
dataSpell$Spell_Count[dataSpell[, TimeVar] >= temp[temp$spell_ID ==
u, TimeVar]] <- u
}
}
return(dataSpell$Spell_Count)
}
Try the DataCombine package in your browser
Any scripts or data that you put into this service are public.
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.