• Monday: Theory
• Warm-up Activity (10 min)
• Discussion on readings (40 min)
• Research design work (25 min)
• Project questions (remainder)

• Wednesday: Laboratory
• Warm-up Activity (10 min)
• New material + discussions (40 min)
• Additional material + discussions (25 min)
• Problem set / final project questions (remainder)

• Intros
• Thinking like data scientists
• Activity
• Intro to R, RStudio, Python, Anaconda, Jupyter Notebooks, and Julia
• Interactive Coding Demo in R
• Installing R and R Studio

• On Zoom
• Syllabus
• Intro to coding in R

• Office Hours: Fridays, 1:30pm-3:00pm

• Name
• Program
• One reason taking this class
• Do you see yourself as a data scientist or social scientist?

• Name
• Program
• One reason taking this class
• Do you see yourself as a data scientist or social scientist?

• In groups

• In groups
• What does Matt Salganik say?

• computation, machine learning, new method
• social importance, poverty

• Far more detailed than previous estimates

• Performed well in accuracy, 10x faster and 50x cheaper than census

• In groups: what are potential benefits?
• In groups: what are potential trade-offs?
• new information!
• could reduce inequality?
• privacy
• accuracy
• generalizability?
• errors
• unknown how it will be used

• Salganik: The best place to start is research design.
• Research question (e.g. what is poverty in Rwandan neighborhoods?)
• Data (e.g. cell phone data)

• Task: design a research study to estimate the number of self-described “data scientists” and “social scientists” at Stanford
• What is the outcome of interest and/or research question?
• What data will you use? (ready made vs. custom made)
• Be creative!

• Ready-Made Data: LinkedIn, github, etc.
• Custom-Made Data: Surveys, compilations of student records, etc.
• Things to consider:
• What can we infer from online information or class schedules?
• Can people be both? neither?
• How costly are our methods?
• Privacy/ethics concerns

• Developed by statisticians
• Lots of resources
• Flexible
• Amazing graphics
• Open source, free
• Syntax challenges from different packages

• Environment for working in R

• Developed by computer scientists
• Lots of resources
• Flexible and more popular than R
• Open source, free
• Similar to R, relies on various packages

• Environment for launching multiple programs, including Jupyter Notebooks

• Flexible and intuitive way to integrate code and text
• Similar to R Markdown in function

• Creates nice pdf, html, documents and more
• Recommended way to turn in assignments for this class (generate pdf or html files)

• Quickly becoming popular in data science! Why?

• tinyurl.com/soc10week1

• Install R and RStudio
• Optional: watch video linked at the end of exercise (~14min)
• Optional: look at syllabus and problem set 1 (linked on canvas)
• See you Wednesday on Zoom!

• Syllabus (20 min)
• Break (5 min)
• Coding! Coding! Coding! (50 min)
• Introduce Problem Set (Remainder)

https://019c932f-6501-a8d5-bac5-df7f6ede5a76.share.connect.posit.cloud/

• Developed by statisticians
• Lots of resources
• Flexible
• Amazing graphics
• Open source, free
• Syntax challenges from different packages

• Environment for working in R

#####################################################
## title: a new R script!
## author: you!
## purpose: to try out R
## date: today's date
#####################################################

# you can start coding below

# what does R do with numbers?
2

## [1] 2

# what if we try adding or multiplying?
4+3

## [1] 7

53*2

## [1] 106

# what if we put all the previous results together?
c(2, 3, 4+3, 53*2, 90*4/5)

## [1]   2   3   7 106  72

• Numeric
• Character
• Logical
• Integer

• Pick a number! Try running this number as different types of data (as.integer, as.numeric, as.character, as.logical)
• Try using conditional statements (<, >, ==, !=, >=, <=) with different character and numeric values

# what if we put all the previous results together?
matrix(c(2, 3, 4+3, 53*2, 90*4/5),
       nrow = 5)

##      [,1]
## [1,]    2
## [2,]    3
## [3,]    7
## [4,]  106
## [5,]   72

• The ? sends you to the help page of whatever function follows it
• Try it! Type ?matrix and then hit ctrl + enter

• Try creating your own matrix!
• Use the matrix() function
• Add numbers by putting c(your numbers here) inside
• Can you create a \(3 \cdot 3\) matrix?
• Note: ? will give you more info on how a function works (i.e. ?matrix)

list(c(2, 3, 4+3),
     53*2,
     90*4/5)

## [[1]]
## [1] 2 3 7
## 
## [[2]]
## [1] 106
## 
## [[3]]
## [1] 72

• Vectors
• Matrices
• Lists

• An important part of coding is assigning values to labels
• We can assign vectors, matrices, lists, and more
• This is helpful for storing them and using them later

# assign "sum" to the sum of a few numbers
penguin <- c(2, 3, 4+3, 53*2, 90*4/5)

# now take a look at the result
penguin

## [1]   2   3   7 106  72

• Something else we can do is apply functions to values

# let's have R sum the numbers
sum(c(2, 3, 4+3, 53*2, 90*4/5))

## [1] 190

• Take a minute to

1. Assign a series of numbers to a label
2. Sum this label

• We can also write our own functions!

# we'll call our function 'addition'
addition <- function(x, y){
  return(x+y)
}

addition(2, 2)

## [1] 4

• As an exercise, try writing your own function
• Start with just one argument (function(x))
• Then try adding more args (function(x,y))

• Loops iterate the same procedure through a series of values

# try printing each of the first 10 numbers

for(i in 1:10){
  print(i)
}

## [1] 1
## [1] 2
## [1] 3
## [1] 4
## [1] 5
## [1] 6
## [1] 7
## [1] 8
## [1] 9
## [1] 10

• As an exercise, try writing your own loop
• What happens if you use something other than i?
• What happens if you use something other than 1:10?

# vector with 1-10
ten <- c(1:10)

# try adding one
ten + 1

##  [1]  2  3  4  5  6  7  8  9 10 11

# vector with 1-10
ten <- c(1:10)

sapply(ten, function(x) x + 1)

##  [1]  2  3  4  5  6  7  8  9 10 11

• A note on coding: we are speaking a language! Communication is key
• Meaning: use lots of #’s

source

• Data science can be done with small data, and limited resources
• What is important is that the analyst thinks carefully about decisions and message

• Detailed Map
• An more modern visualization

• Could we do the following by hand?