In JayAchar/tbcleanr: tbcleanr: tools for cleaning TB datade

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

The tbcleanr package has been designed to simplify routine data cleaning for TB programmatic data. Since this data is typically stored in Excel, CSV or TSV formatted files, the chosen workflow for the package retains this seperation to allow easier calculation and generation of new variables in the tbgeneratr package. In this vignette, a typical workflow for cleaning admission data files is presented.

Import data

Use of the readr package is recommended for importing CSV and TSV files. While the readxl package can be used for data stored in Excel documents, this is not recommended.

# define working directory
setwd("C:/tb_data")

# confirm required admission file is present
list.files()

# read in admission data
adm <- readr::read_csv("admission_tb_data.csv", guess_max = 100000)

The guess_max option is defined here to ensure variables are parsed using a greater than standard figure. Since some variables have been added to data sets only in recent years, if this option is not defined, there is a risk that recently collected data will be incorrectly lost through misclassification of a variable as a logical vector. For larger data sets, higher figures for guess_max are recommended.

Check for warnings in the output of the read_csv() call. If a variable has been incorrectly defined as a logical, parsing failures will be displayed.

Simplified workflow

All tbcleanr functions designed to work with admission data files can be applied using one function call:

clean_adm <- adm_data_cleanr(adm)

Individual functions can also be called seperately where more specific cleaning is required.

Multiple files

Sometimes cleaning and subsequent analysis of multiple data sets is required. This can be achieved by combining the tbcleanr and tbgeneratr packages with purrr.

``` {r multiple_files, include = TRUE, eval = FALSE}

define files to process

filenames <- c("admission1.csv", "admission2.csv", "admission3.csv")

combine purrr with tbcleanr

tbl <- dplyr::tibble(file = filenames, adm_data = map(.x = filenames, .f = ~ {readr::read_csv(.x, guess_max = 100000) %>% tbcleanr::adm_data_cleanr()})) ```

This approach uses map() to iterate over the defined file names in the working directory. The defined function within map() imports the file and applies the adm_data_cleanr() leaving a cleaned, nested dataset in the tibble. Further use of map() will allow additional processing using the tbgeneratr package.

JayAchar/tbcleanr documentation built on Aug. 12, 2020, 8:40 a.m.