makeCohort: makeCohort() takes a vector of CPS2 variable names and merges...

In buddha2490/MargotFun: Just a bunch of useful functions to do my job.

Description

This function will pull all the variables listed in an input vector, pull these variables from each of the CPS2, Nutrition Cohort, and Lifelink and merge them into a single file for later processing. It is a simple program, just saves me from having to write a lot of readRDS() expressions.

Note: The returned data will be subset to MASTER==1 and NUT_92==1.

Usage

makeCohort(vars, sex = "BOTH", intialPath = file.path("s:/cps"))

Arguments

vars	Character vector of variable names. Do not include "ID" in the list. Not case sensitive.
sex	"MEN", "WOMEN" or "BOTH". Defaults to "BOTH"
intialPath	Defaults to the S: drive, but I can also put in a local path.

Value

A data frame including all the variables listed in the input vector.

Author(s)

Brian Carter

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (vars, sex = "BOTH", intialPath = file.path("s:/cps"))
{
    require(dplyr)
    message("Last updated on July 30th 2019")
    message("Cohort only includes data 1982, and 1992-2015.  If followup goes later, please update the function")
    menFiles <- NULL
    menVars <- NULL
    womenFiles <- NULL
    womenVars <- NULL
    if (toupper(sex) == "MEN") {
        foo <- dplyr::filter(variables$men, Variables %in% vars)
        foo$FullPath <- file.path(initialPath, foo$subPath, foo$File)
        menFiles <- unique(foo$FullPath)
        menVars <- c(unique(foo$Variables), "ID")
    }
    if (toupper(sex) == "WOMEN") {
        foo <- dplyr::filter(variables$women, Variables %in%
            vars)
        foo$FullPath <- file.path(initialPath, foo$subPath, foo$File)
        womenFiles <- unique(foo$FullPath)
        womenVars <- c(unique(foo$Variables), "ID")
    }
    if (toupper(sex) == "BOTH") {
        foo <- dplyr::filter(rbind(variables$men, variables$women),
            Variables %in% vars)
        foo$FullPath <- file.path(initialPath, foo$subPath, foo$File)
        menFiles <- unique(foo$FullPath[foo$Sex == "MEN"])
        menVars <- c(unique(foo$Variables[foo$Sex == "MEN"]),
            "ID")
        womenFiles <- unique(foo$FullPath[foo$Sex == "WOMEN"])
        womenVars <- c(unique(foo$Variables[foo$Sex == "WOMEN"]),
            "ID")
    }
    head(foo)
    cl <- parallel::makeCluster(parallel::detectCores(logical = T))
    parallel::clusterExport(cl, c("foo", "menFiles", "menVars",
        "womenFiles", "womenVars"))
    menCohort <- parallel::parLapply(cl, menFiles, function(x) {
        dat <- readRDS(x)
        dat <- dat[, names(dat) %in% menVars]
        return(dat)
    }) %>% Reduce(function(x, y) full_join(x, y, "ID"), .)
    womenCohort <- parallel::parLapply(cl, womenFiles, function(x) {
        dat <- readRDS(x)
        dat <- dat[, names(dat) %in% womenVars]
        return(dat)
    }) %>% Reduce(function(x, y) full_join(x, y, "ID"), .)
    parallel::stopCluster(cl)
    rm(cl)
    if (toupper(sex) == "MEN") {
        return(menCohort)
    }
    if (toupper(sex) == "WOMEN") {
        return(womenCohort)
    }
    if (toupper(sex) == "BOTH") {
        return(dplyr::bind_rows(menCohort, womenCohort))
    }
  }

buddha2490/MargotFun documentation built on Nov. 4, 2019, 8:16 a.m.