keyImport: Import/validate a key object or import/validate a key from a...

In kutils: Project Management Tools

Import/validate a key object or import/validate a key from a file.

Description

After the researcher has updated the key by filling in new names and values, we import that key file. This function can import the file by its name, after deducing the file type from the suffix, or it can receive a key object from memory.

Usage

keyImport(
  key,
  ignoreCase = TRUE,
  sep = c(character = "\\|", logical = "\\|", integer = "\\|", factor = "\\|",
    ordered = "[\\|<]", numeric = "\\|"),
  na.strings = c("\\.", "", "\\s+", "N/A"),
  missSymbol = ".",
  ...,
  keynames = NULL
)

Arguments

key	A key object (class key or keylong) or a file name character string (ending in csv, xlsx or rds).
ignoreCase	In the use of this key, should we ignore differences in capitalization of the "name_old" variable? Sometimes there are inadvertent misspellings due to changes in capitalization. Columns named "var01" and "Var01" and "VAR01" probably should receive the same treatment, even if the key has name_old equal to "Var01".
sep	Character separator in value_old and value_new strings in a wide key. Default is are "|". It is also allowed to use "<" for ordered variables. Use regular expressions in supplying separator values.
na.strings	Values that should be converted to missing data. This is relevant in name_new as well as value_new. In spreadsheet cells, we treat "empty" cells (the string ""), or values like "." or "N/A", as missing with defaults ".", "", "\s" (white space), and "N/A". Change that if those are not to be treated as missings.
missSymbol	Defaults to period "." as missing value indicator.
...	additional arguments for read.csv or read.xlsx.
keynames	Don't use this unless you are very careful. In our current scheme, the column names in a key should be c("name_old", "name_new", "class_old", "class_new", "value_old", "value_new", "missings", "recodes"). If your key does not use those column names, it is necessary to provide keynames in a format "our_name"="your_name". For example, keynames = c(name_old = "oldvar", name_new = "newname", class_old = "vartype", class_new = "class", value_old = "score", value_new = "val").

Details

This can be either a wide or long format key file.

This cleans up variables in following ways. 1) name_old and name_new have leading and trailing spaces removed 2) value_old and value_new have leading and trailing spaces removed, and if they are empty or blank spaces, then new values are set as NA.

Policy change concerning empty "value_new" cells in input keys (20170929).

There is confusion about what ought to happen in a wide key when the user leaves value_new as empty or missing. Literally, this means all values are converted to missing, which does not seem reasonable. Hence, when a key is wide, and value_new is one of the na.strings elements, we assume the value_new is to be copied from value_old. That is to say, if value_new is not supplied, the values remain same as in old data.

In a long key, the behavior is different. Since the user can specify each value for a variable in a separate row, the na.strings appearing in value_new are treated as missing scores in the new data set to be created.

Value

key object, should be same "wide" or "long" as the input Missing symbols in value_old and value_new converted to ".".

Author(s)

Paul Johnson <[e-mail]>

Examples

mydf.key.path <- system.file("extdata", "mydf.key.csv", package = "kutils")
mydf.key <-  keyImport(mydf.key.path)
## Create some dupes
mydf.key <- rbind(mydf.key, mydf.key[c(1,7), ])
mydf.key2 <- keyImport(mydf.key)
mydf.key2
## create some empty value_new cells
mydf.key[c(3, 5, 7) , "value_new"] <- ""
mydf.key3 <- keyImport(mydf.key)
mydf.key3
mydf.keylong.path <- system.file("extdata", "mydf.key_long.csv", package = "kutils")
mydf.keylong <- keyImport(mydf.keylong.path)

## testDF is a slightly more elaborate version created for unit testing:
testdf.path <- system.file("extdata", "testDF.csv", package = "kutils")
testdf <- read.csv(testdf.path, header = TRUE)
keytemp <- keyTemplate(testdf, long = TRUE)
## A "hand edited key file"
keyPath <- system.file("extdata", "testDF-key.csv", package="kutils")
key <- keyImport(keyPath)
keydiff <- keyDiff(keytemp, key)
key2 <- rbind(key, keydiff$neworaltered)
key2 <- unique(key)
if(interactive())View(key2)

kutils documentation built on Sept. 17, 2023, 5:06 p.m.