• Instructor: Sotirios Damouras
  • Pronounced s-aw-t-EE-r-ee-aw-s or Sam
  • Email: sdamouras@utsc.utoronto.ca
  • Webpage: http://www.utsc.utoronto.ca/~sdamouras/
  • Office hours: MO 11-13 / WE 11-13 / FR 13-15 @ IC456
• TA's
  • Xichen (Sherry) He (xichen.he@mail.utoronto.ca)
  • Jiali Pan (jiali.pan@mail.utoronto.ca)
  • Xincheng Zhang (xincheng.zhang@mail.utoronto.ca)

• Science: systematic organization & creation of knowledge, in form of testable explanations/predictions about the world

• Data Science: field that uses scientific methods to extract knowledge & insights from data
  • Scientific method invloves making conjectures and testing them using observation or experimentation.

• In short, this course is about answering questions using data.

• Unfortunately, cannot answer every question using data.

• But certain types of questions can be reasonably addressed with data
  • Estimation: What % of the population gets the flu?
  • Inference: Should I take drug X for the flu?
  • Prediction: Will I get the flu?

• Be able to conduct a systematic investigation of a question/problem using data. In particular:
  • Organize & manipulate data
  • Explore & investigate data graphically
  • Deal with variability & estimation
  • Formulate & test statistical hypotheses
  • Discover patterns in data & use them for prediction
  • Develop strong statistical computing skills
  • Communicate statistical ideas & results effectively

• Logistics
  • All course material (lecture slides, assignments, grades, communications) available through Quercus
• Textbooks
  • R for Data Science
  • Introductory Statistics with Randomization and Simulation
  • An Introduction to Statistical Learning

• Project Topic Is university education worth it ?

• Purposefully open-ended question; answer various aspects using data
  • Collect and examine relevant data
  • Formulate interesting questions
  • Use data to address them convincingly

• Hope you don't drop out based on your analysis :)

• Getting started with R/RStudio
  • Using the RStudio IDE
  • Quick run-through of R
  • Writing RMarkdown reports
• Readings
  • Quick guide to Rmarkdown

• Install R and RStudio for Linux/Windows/Mac
  • Alternatively, use cloud version at RStudio Cloud

• Create folder and associated Rstudio project
  • Container for your data, code, and output files

• Write code (.R/Rmd scripts) in Editor
• Run commands/scripts, interact with R in Console
• Check available variables/data in Environment
• Look at output/documentation/files in Viewer

• Information is contained in objects
  • E.g. data, variables, models, plots
• Operations are performed by functions
  • E.g. sorting data, fitting models, plotting results
• Carry out analysis by applying functions to objects

my_numbers = 1:3
sum(my_numbers)
## [1] 6

• R comes pre-loaded with basic functions; for extra functionality there are two options:
  • Create your own: R programming
  • Use someone else's: R Packages
• R Packages are bundles of reusable functions, data, & documentation
  • Packages must be downloaded once w/ install.packages()
  • To use a package, you must load it into R session w/ library()

install.packages("package_name")
library(package_name)

• Get function documentation using help() or ?
  • See description, arguments, results & examples

help(fun_name)  
? help 

• For help on packages

help( package = "package_name" )

• Read/write tabular data from/to spreadsheet-like file (.csv or comma separated values)

my_data = read.csv( file = "C:/Users/Sotiris/Documents/data.csv")
write.csv( my_data, file = "C:/Users/Sotiris/Documents/data.csv")

• Read/write binary representation of select R object

save( obj1, obj2, file = "C:/Users/Sotiris/Documents/my_objects.Rdata")

• Save all objects in in .RData file

save.image()

• Absolute file paths ('C:/Users/Sotiris/…') are BAD for reproducibility

• Instead, use relative paths with respect to your working directory
  • Find/set working directory with getwd()/setwd()
  • Access working directory directly with

getwd()
my_data = read.csv( file = 'data_folder/data.csv')

• Rmarkdown is a framework for creating reproducible reports

• Rmarkdown scripts combine together:
  • R code for performing data analysis
  • Markdown code for authoring documents

• Actually, these very slides are written in Rmarkdown

• Let's create your first report with Rmarkdown