Steph Locke

• Locke Data
• Presenter on Data Science, BI, and DataOps
• Microsoft Data Platform MVP
• User group leader: MSFT Stack & CaRdiff
• Conference organiser: satRdays
• Author: R Fundamentals

github.com/StephLocke | itsalocke.com | @theStephLocke

Agenda

• What is R?
• R for data manipulation
• R for data visualisation
• R for reporting
• ~R for data science~
• Nifty stuff
• Next steps

R

• What is it?
• Origins
• Quirks
• Value

Use case - wrangling data

library(tidyverse, quietly = TRUE)
head(population, 2)

## # A tibble: 2 x 3
##   country      year population
##   <chr>       <int>      <int>
## 1 Afghanistan  1995   17586073
## 2 Afghanistan  1996   18415307

head(spread(population, year, population), 2)

## # A tibble: 2 x 20
##   country   `1995` `1996` `1997` `1998` `1999` `2000` `2001` `2002` `2003`
##   <chr>      <int>  <int>  <int>  <int>  <int>  <int>  <int>  <int>  <int>
## 1 Afghani~  1.76e7 1.84e7 1.90e7 1.95e7 2.00e7 2.06e7 2.13e7 2.22e7 2.31e7
## 2 Albania   3.36e6 3.34e6 3.33e6 3.33e6 3.32e6 3.30e6 3.29e6 3.26e6 3.24e6
## # ... with 10 more variables: `2004` <int>, `2005` <int>, `2006` <int>,
## #   `2007` <int>, `2008` <int>, `2009` <int>, `2010` <int>, `2011` <int>,
## #   `2012` <int>, `2013` <int>

Use case - visualisation

library(datasauRus)
ggplot(datasaurus_dozen, aes(x, y)) + geom_point() + facet_wrap(~dataset, 
    ncol = 5)

Use case - integration

library(reticulate)
os <- import("os")
os$getcwd()

## [1] "T:\\sites\\superbuild\\pres-r\\pres"

Tidyverse

• Lots of data sources and destinations
• SQL-ish functions
• Kicks SQLs butt
• “Piping” for easy to read code

I/O packages

• readr for flat files
• readxl for Excel
• haven for SAS, Stata, etc
• odbc for databases
• jsonlite for JSON
• xml2 for XML
• httr for APIs
• rvest for web scraping
• sparklyr for Spark

Data manipulation packages

• dplyr for “SQL”
• dbplyr for “SQL” against DBs
• tidyr for pivoting and unpivoting data
• purrr for loop(ish) stuff without the loops
• magrittr for pipelines

Bulk read example

list.files("../demo/multiread", full.names = TRUE) %>% map_df(read_csv, 
    col_types = cols(.default = col_character()))

## # A tibble: 105 x 7
##    Timestamp  `Your first name` `Your favourite i~ `Your pet's name (with~
##    <chr>      <chr>             <chr>              <chr>                  
##  1 15/06/201~ Maëlle            Dulce de leche     <NA>                   
##  2 15/06/201~ Dave              Lemon              Cally                  
##  3 15/06/201~ Ahmadou           Caramel            Tigrou                 
##  4 15/06/201~ Arun              Strawberry         Zeus                   
##  5 15/06/201~ Ed                Peach              Charlie                
##  6 15/06/201~ Kyle              Blueberry          Nemo                   
##  7 15/06/201~ Lilian            raspberry          <NA>                   
##  8 15/06/201~ Bea               lemon              peppers                
##  9 15/06/201~ Fred              cookie dough       Max                    
## 10 15/06/201~ David             Vanilla            Jesus                  
## # ... with 95 more rows, and 3 more variables: `The number of
## #   floor/stories your residence has` <chr>, `The year you were born in
## #   (YYYY)` <chr>, `Your favourite day of the week` <chr>

Data pipelines

starwars %>% unite(homeworld_species, homeworld, species) %>% 
    count(homeworld_species)

## # A tibble: 58 x 2
##    homeworld_species            n
##    <chr>                    <int>
##  1 Alderaan_Human               3
##  2 Aleen Minor_Aleena           1
##  3 Bespin_Human                 1
##  4 Bestine IV_Human             1
##  5 Cato Neimoidia_Neimodian     1
##  6 Cerea_Cerean                 1
##  7 Champala_Chagrian            1
##  8 Chandrila_Human              1
##  9 Concord Dawn_Human           1
## 10 Corellia_Human               2
## # ... with 48 more rows

Databases

library(DBI)
library(odbc)

driver = "ODBC Driver 13 for SQL Server"
server = "mhknbn2kdz.database.windows.net"
database = "AdventureWorks2012"
uid = "sqlfamily"
pwd = "sqlf@m1ly"

dbConn <- dbConnect(odbc(), driver = driver, server = server, 
    database = database, uid = uid, pwd = pwd)

Querying databases

library(dbplyr)

medalist <- tbl(dbConn, in_schema("olympics", "medalist"))
sport <- tbl(dbConn, in_schema("olympics", "sport"))

medalist %>% left_join(sport)

## Joining, by = "SportID"

## # Source:   lazy query [?? x 12]
## # Database: Microsoft SQL Server
## #   12.00.0600[sqlfamily@mhknbn2kdz/AdventureWorks2012]
##    MedalistID Event  Edition Athlete  NOC   Gender Medal Season MedalValue
##         <int> <chr>    <int> <chr>    <chr> <chr>  <chr> <chr>       <dbl>
##  1          1 1500m~    1900 HALMAY,~ HUN   Men    Bron~ Summer          3
##  2          2 1500m~    1900 JARVIS,~ GBR   Men    Gold  Summer          1
##  3          3 1500m~    1900 WAHLE, ~ AUT   Men    Silv~ Summer          2
##  4          4 200m ~    1900 DROST, ~ NED   Men    Bron~ Summer          3
##  5          5 200m ~    1900 HOPPENB~ GER   Men    Gold  Summer          1
##  6          6 200m ~    1900 RUBERL,~ AUT   Men    Silv~ Summer          2
##  7          7 200m ~    1900 HALMAY,~ HUN   Men    Silv~ Summer          2
##  8          8 200m ~    1900 RUBERL,~ AUT   Men    Bron~ Summer          3
##  9          9 200m ~    1900 LANE, F~ AUS   Men    Gold  Summer          1
## 10         10 200m ~    1900 KEMP, P~ GBR   Men    Bron~ Summer          3
## # ... with more rows, and 3 more variables: DisciplineID <int>,
## #   SportID <int>, Sport <chr>

Key packages

• ggplot2 for static viz
  • ggthemes for extra themes
  • ggraph for graph viz in ggplot
  • ggmap for geospatial in ggplot
• plotly for interactive viz
• htmlwidgets for generating anything html / javascript
• DiagrammeR for flowcharts and things

ggplot2 basics

ggplot(datasaurus_dozen, aes(x, y)) + geom_point() + facet_wrap(~dataset, 
    ncol = 5)

Turning a ggplot interactive

library(plotly, quietly = TRUE)
p <- ggplot(datasaurus_dozen, aes(x, y)) + geom_point() + facet_wrap(~dataset, 
    ncol = 5)
ggplotly(p)

## We recommend that you use the dev version of ggplot2 with `ggplotly()`
## Install it with: `devtools::install_github('hadley/ggplot2')`

Building interactive plots

map_data("world", "UK") %>% group_by(group) %>% plot_geo(x = ~long, 
    y = ~lat) %>% add_markers(size = I(1))

Key packages

• rmarkdown for using markdown
• knitr for converting R markdown files to other formats
• revealjs for whizzy presentations
• bookdown for making books
• blogdown for sites
• officer for complex(ish) Word and PowerPoint
• shiny for interactive reports
• flexdashboard for dashboards
• DataExplorer for quick reports on data

rmarkdown & revealjs

See slide source code!

bookdown

See Data Manipulation in R source code

shiny & flexdashboard

See Cost of Coffee app & source code

DataExplorer

DataExplorer::create_report(who)

Plays well with others

• Consume Python & JS
• Consume from Python & JS
• Incoporated into many BI platforms
• Run in-database
• Run in Spark

Community

• R UGs
• R Ladies
• Conferences e.g. SatRday Cardiff June 23rd
• Twitter (#rstats)
• r-bloggers.com
• r-weekly.com

Q&A

Learn

• Community
• datacamp.com
• My books

Follow up with me

• @theStephLocke
• Join my #learnr slack! bit.ly/ldlearnrslack
• Get slides and blog posts itsalocke.com