README.md

In joepowers16/breadcrumbs: Breadcrumbs Helps You Maintain A Tidy Workflow During EDA

Who should you use breadcrumbs?

2019-12-20

If you hate being unable to find code and data in old data projects, you may find breadcrumbs to be a breath of fresh air.

The goal for the breadcrumbs package is to provide a routine structure to R projects so that you put code and data in the right places and find them later with ease. breadcrumbs also encourages you to coordinate your code and data with Makefile(s), which are a wonderful resource for reproducibile workflows.

Getting started

To get started using breadcrumbs 1. open a new or existing R project 2. Load the breadcrumbs package library(breadcrumbs) 2. Execute the create_project_directories() function in your console

You will now see a standard set of subdirectories appear in your project directory.

The subdirectory structure

<project>_cloud_dir/ I use the <project>_cloud_dir/ to save large files like data and image-laden reports that don’t belong in GitHub. I recommend you sync this folder to the cloud via Google Drive or Box so that it is backed up off your local machine. ….reports/ to store finished reports and presentations ….data/ to store processed data ……..raw/ to store raw data R/ to store your functions munge/ to store scripts that output processed data analysis/ to store your analysis scripts

breadcrumbs is filled with functions to support the use of these subdirectories. For instance, I have found it useful to save my filepaths to commonly used subdirectories in a dedicated script called “file_paths.R”. The function source_file_paths() will execute this script, and I include this function at the top of all my other scripts and Rmd files.

Makefiles

create_project_directories() has also added a Makefile to your project’s home directory. This Makefile can be used to make your entire project reproducible from raw data to finished reports. For more details on how to use Makefiles see the appendix in the Makefile template generated by breadcrumbs.

Other useful functions

wrap()

In addition to directory structure, breadcrumbs includes a number of functions I find useful at the exploratory stage of data projects. For instance wrap() will control the width of long ggplot titles and facilitate your writing accurate and informative titles on all your EDA plots without fussing around trying to find the right places to manually insert line breaks.

library(breadcrumbs)
library(tidyverse)

ggplot(mtcars, aes(x = wt, y = mpg)) + 
    geom_point() + 
    labs(title = wrap("The wrap function frees me up to write long and informative titles on my plots without having to troubleshoot the placement of linebreaks. This is useful since I often forget what I once found useful about a plot.", 80)) + 
    theme_minimal()

clean_names()

Another useful function is clean_names(). I use clean_names() the first time I import raw data to remove capital letters and illegal characters. I’ll read in a government dataset below to demonstrate. Note how illegal names like JURISDICTION NAME become clean snake_case names like jurisdiction_name. This is a great step to take at the start of your project so you don’t need to remember esoteric conventions about how variables were names in different datasets.

file_nyc_data <- 
    "https://data.cityofnewyork.us/api/views/kku6-nxdu/rows.csv"

# Reading data without clean_names()
read_csv(file_nyc_data) %>% select(1:4)

## # A tibble: 236 x 4
##    `JURISDICTION NAME` `COUNT PARTICIPANTS` `COUNT FEMALE` `PERCENT FEMALE`
##                  <dbl>                <dbl>          <dbl>            <dbl>
##  1               10001                   44             22             0.5 
##  2               10002                   35             19             0.54
##  3               10003                    1              1             1   
##  4               10004                    0              0             0   
##  5               10005                    2              2             1   
##  6               10006                    6              2             0.33
##  7               10007                    1              0             0   
##  8               10009                    2              0             0   
##  9               10010                    0              0             0   
## 10               10011                    3              2             0.67
## # … with 226 more rows

# Reading data with clean_names()
read_csv(file_nyc_data) %>% select(1:4) %>% clean_names()

## # A tibble: 236 x 4
##    jurisdiction_name count_participants count_female percent_female
##                <dbl>              <dbl>        <dbl>          <dbl>
##  1             10001                 44           22           0.5 
##  2             10002                 35           19           0.54
##  3             10003                  1            1           1   
##  4             10004                  0            0           0   
##  5             10005                  2            2           1   
##  6             10006                  6            2           0.33
##  7             10007                  1            0           0   
##  8             10009                  2            0           0   
##  9             10010                  0            0           0   
## 10             10011                  3            2           0.67
## # … with 226 more rows

joepowers16/breadcrumbs documentation built on Dec. 25, 2019, 4:13 p.m.