gistfile1.md

Introduction to R for Hedonic Regression

1. Setting the stage
2. Basics of the language
3. Demo in RStudio
4. Strategies for writing code

1. Setting the stage

What is R?

• Public-domain programming language for statistics

Why write code?

• Precise and replicable
• Efficient and powerful for large datasets

Comparing tools

• There's a range of statistics tools from full programming languages to spreadsheets
• Python -> R -> Stata -> Excel
• General tradeoff of flexibility vs. ease of use
• Some are public domain, some are commercial ($$)
• Different specializations, different communities of users

2. Basics of the language

Basics

• Calculations

3 * 4 + 6

• Variables and assignment

a = 3 * 4 + 6
a <- 3 * 4 + 6
a<-3*4+6
b = a / 3
b = c / 3
b = b ^ 2

Functions

• Syntax, arguments, return values

c = round(5.27)
bigNumber = exp(200)
big.number = exp(200)
bigNumber2 = exp(200)

• Optional arguments

c = round(5.27, digits=1)
c = round(5.27, 1)

• Nested functions

c = round(sqrt(2) + 1)

• Side effects

print("Arrr")

Data types

• Integer, numeric (a.k.a. real number), logical (boolean), character (string)

city = "Berkeley"
address = paste("College Ave, ", city)
print(typeof(city))

• Vector (a.k.a. array)

vec = c(10, 20, 30, 40, 50)
vec[2]
vec[1:3]

• Data frame = matrix with named columns

Control structures

• If/then statement

if (rent > 1000) {
print("too expensive")
}

• Loop

for (i in 1:10) {
print(i^2)
}
for (city in c("Berkeley", "Oakland", "San Francisco")) {
print(paste(city, " is the best city"))
}

Where do functions come from?

• Some are built in and some are part of packages you can install
• You can write your own

3. Demo in RStudio

Orientation

• code window, console, data window, documentation window

Demo walkthrough

• downloading packages
• reading in data
• subsetting data frames
• descriptive statistics
• histogram, scatter plot
• generating new variables
• linear regression
• reading regression output

Pitfalls

• avoid special characters and keywords
• missing values
• CSV formatting errors
• incorrect data type inference

4. Strategies for writing code

How does this function work??

• Search the reference documentation
• Focus on required arguments, optional arguments, return values

What other functions exist??

• Look through narrative documentation like the "Short Intro" document
• Search for the concept (like "R percentile") in reference documentation or online

How to decipher code that someone else wrote

• Trace through the logic line by line
• Pay special attention to variable creation and to function calls

Diagnosing problems

• Error messages vs. unexpected results
• Most problems are caused by typos or unanticipated data values (wrong type, zero, too large, missing, etc)
• For unexpected results, check the variable contents after each line of code to isolate the problem

Dealing with error messages

• How to interpret them (see "Short Intro" document)
• Find which line generates the error
• Look for syntax mistakes on that line
• If there are no obvious mistakes, trace backward through the logic that created each of the variables used on that line

Google, Google, Google

• ...but pay attention to which code resource websites are helpful and which are spammy
• Stack Overflow is excellent
• stat.ethz.ch is official R documentation