R

class: center, middle, inverse, title-slide

# R
## What is it and why should you care?
### Steve Crawshaw
### 2020-12-25

---

# What is R?

+ R is a programming language designed for statistics and data science

+ It is usually run in a program called R Studio. <https://www.rstudio.com/>

---
# What can it do?

## Data Science

+ Data Analysis
  
--
  
  
+ Data Processing 
  
--
  
  
+ Modelling and machine learning 
  
  
--
  
  
+ Visualisations  
  
  
--
  
  
+ Reporting 
  
--
  
  
+ Web apps (with Shiny) 
  
--
  
  
+ Mapping - like GIS

---
# Why Use R?

## It's Open Source

+ It's free!

+ It's easier to share your data & code

+ Innovations spread quickly

+ *you* can contribute!

---
# Why Use R?

--
## It's powerful & flexible

+ You can use R for more than data analysis, including:

+ creating websites
  
  + creating documents that reproduce your analyses
  
  + slideshows (including *this one*)
  
  + books

+ In R, it is never *if* but *how*

---
# Why Use R?

--
## Reduce Errors, Enhance Reproducibility & Transparency

+ Generate publication quality figures & tables

+ Create detailed and fully documented scripts showing every step between raw data & a statistic

+ You can use R to report your analyses, reducing all too common errors in reported statistics see [see Nuijten et al.](https://link.springer.com/article/10.3758/s13428-015-0664-2).

---
# Why Use R?

--
## It's Efficient
+ Once you get used to it and start using R, it saves you time in the long run.

+ Scripts make re-using past work or using others' work as a starting point much easier.

+ Typing scripts is much faster than clicking through menus, *especially* after you get the hang of keyboard shortcuts

+ It runs faster and is less bloated than GUI stat software (e.g., SPSS)

---
# What are the alternatives?

## Excel

+ Does not separate data from code
+ Limited to 1m lines
+ Slow
+ Very prone to human error
+ See <https://blogs.oracle.com/smb/10-of-the-costliest-spreadsheet-boo-boos-in-history>

--
## Python

+ General purpose language
+ Used for data science
+ R has more libraries (packages) for specialised fields like Air Quality(!)

---

# Alternatives
## Business Intelligence Products

+ Power BI
  + Microsoft - O365
  + BCC uses it
  + Modelling and ML capabilities are limited

+ Tableau
  + Expensive
  + Similar limitations on modelling and ML

---

# But:
## R is difficult to learn!

---
# Examples of using R in BCC
--

+ **Using the Open Data API to publish ratified data**
  + *Extract* - connect to corporate database and National AQ site
  + *Transform* - reshape and process data
  + *Load* - POST data to remote server

+ **<a href="S://SUSTAIN//EnvQual//Air_Quality//Projects//R%20Projects//airquality_GIT//covid-19-air_quality_Bristol_REPORT.html" target="_blank">Regular updates to AQ Board on COVID and air quality</a>**
  + Uses machine learning to "De - weather" air quality measurements
  + Analysis is updated with new data automatically every month
  + Graphics saved as files to drop into PPT

+ **<a href="S://SUSTAIN//EnvQual//Air_Quality//Projects//R%20Projects//tubes//asr_tubes_tables.html" target="_blank">Annual reporting of air quality to Defra (QA process)</a>**
  + Streamline statutory workflow and add QA steps
  + Carries out multiple adjustment calculations and derives KPI's
  + Generates pre - formatted tables to drop into Defra report

+ **<a href="S://SUSTAIN//EnvQual//Air_Quality//Projects//R%20Projects//dashboard_demo//short_report_ods.html" target="_blank">Short report on air quality for public</a>**
  + Summarises AQ data in short format
  + Compelling and interactive graphics with HTML and javascript

---
# Observations and next steps

+ Data science is transforming business

+ Will this be true for local authorities?

+ What insights might be possible with your / our data?
  
--

+ Learning about ML and statistical modelling (thanks!)

---

# **Questions?**