Syllabus - Kosuke Imai's Homepage - Princeton University
Syllabus - Kosuke Imai's Homepage - Princeton University
Syllabus - Kosuke Imai's Homepage - Princeton University
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
POL 245: Visualizing Data<br />
<strong>Syllabus</strong><br />
Summer 2014<br />
Jonathan Olmsted (Instructor)<br />
Bella Wang (Preceptor)<br />
<strong>Kosuke</strong> Imai (Course Head)<br />
Department of Politics, <strong>Princeton</strong> <strong>University</strong><br />
In this course, we consider ways to illustrate compelling stories hidden in a blizzard of data. Equal parts<br />
art, programming, and statistical reasoning, data visualization is a critical tool for anyone doing analysis.<br />
In recent years, data analysis skills have become essential for those pursuing careers in policy advocacy and<br />
evaluation, business consulting and management, or academic research in the fields of education, health,<br />
medicine, and social science. This course introduces students to the powerful R programming language and<br />
the basics of creating data-analytic graphics in R. From there, we use real datasets to explore topics ranging<br />
from network data (like social interactions on Facebook or trade between counties) to geographical data<br />
(like county-level election returns in the US or the spatial distribution of insurgent attacks in Afghanistan).<br />
No prior background in statistics or programming is required or expected.<br />
Contact Information<br />
Name Jonathan Olmsted Bella Wang <strong>Kosuke</strong> Imai<br />
Office Corwin 029 Corwin 026 Corwin 036<br />
Email jolmsted@princeton.edu bxw@princeton.edu kimai@princeton.edu<br />
Office Hours Fri: 2:00pm – 3:30pm Mon: 2:00pm – 3:30pm by appointment<br />
During our office hours we may be in either the office that is listed above or in Corwin 023 (just across<br />
the hall) which has space for more students. If these office hours do not fit your schedule, you should feel<br />
free to contact us directly via email. Also, do not forget about Piazza and QuantLab (see below) where<br />
you can ask questions and receive answers back immediately.<br />
Logistics<br />
Lectures. Monday and Wednesday, 9:00am–9:50am, Sherrerd Hall 101. Lecture slides will be posted on<br />
Blackboard by 10:00pm the night before the lecture. Students should bring their own printed copy of the<br />
slides to the lecture. The schedule during the first week deviates from this, details are below in the Course<br />
Outline section.<br />
Precepts. Tuesday and Thursday, 9:00am–10:20am, Friend Center 005 and 007. These are computer<br />
labs, and so there is no need to bring a personal laptop (although it is fine for you to do so). The schedule<br />
during the first week deviates from this, details are below in the Course Outline section.<br />
1
Review Sessions/Guest Speakers. Friday, 9:00am–10:20am, Visualization Laboratory in Lewis Library<br />
346. These sessions occur during the second week of FSI through the final week. The first four of<br />
these five sessions involve guest speakers from various industries where data visualization is used. The<br />
schedule during the first week deviates from this, details are below in the Course Outline section.<br />
Lunch with Guest Speaker. Friday, 12:00pm–1:30pm, Prospect House. Students will sign up to have<br />
lunch with one of the four guest speakers at the beginning of the course. During the selected week, students<br />
and the course team will meet with the guest speaker during a casual, catered lunch. The schedule during<br />
the first week deviates from this, details are below in the Course Outline section.<br />
Course Requirements<br />
• Precept participation (15%): There will be twelve precepts in the computer lab. Students must<br />
actively participate in these precepts. During each of the six weeks, there will be an assignment due<br />
before the first precept of that week. This assignment asks students to work on a small set of Precept<br />
Questions and submit their answers the night before the first precept. Details on these assignments<br />
are below. This is an individual assessment with limited collaboration.<br />
• Problem sets (40%): There will be four problem sets during the semester. These problem sets<br />
provide an opportunity for students to conduct analysis and visualization of a particular kind according<br />
to the week’s material. Each problem set will be equally weighted. Details on these assignments<br />
are below. This is an individual assessment with no collaboration.<br />
• Midterm Exam (25%): There will be a take-home midterm exam after the first three weeks of<br />
the course. This exam will provide students an opportunity to demonstrate proficiency of the introductory<br />
material that the subsequent topical weeks will build on. This is an individual assessment<br />
with no collaboration.<br />
• Final Project (20%): The final evaluation is a group data analysis project. Students will be<br />
assigned to groups. Each group will be assigned a dataset from those provided by the four guest<br />
speakers throughout the course. Using the assigned dataset, students will give a 5 minute presentation<br />
to the class focused on a compelling relationship or story from the dataset using no more than 3<br />
figures. A full set of details regarding the final project will be provided later in the course. This is<br />
a group assessment with collaboration allowed only within the assigned groups.<br />
Submission of Computer Code via Blackboard Folders<br />
For the Precept Questions on the Precept Handouts, answers to Problem Sets, and answers to<br />
the Midterm exam, students are required to turn in their computer code via assignment folders on<br />
Blackboard. For each assignment, you will submit your code to the appropriate folder as a single file<br />
named xxxHandoutX.R or xxxPSetX.R where xxx is your NetID and X is the handout/problem set number.<br />
For example, it might be kimaiHandout2.R or jolmstedPSet3.R for handout 2 or problem set 3.<br />
Precept Materials, Format, and Policy<br />
Precept Handouts. We will distribute a Precept Handout as preparation for the first precept of<br />
each week. It contains all the materials we expect you to learn before that session. Mastery of these<br />
materials is required for Problem Sets, Precept Worksheets, and the Midterm Exam. These<br />
2
handouts will be posted on Friday afternoon and due the night before Tuesday’s precept. You will want<br />
paper copies of Precept Handouts to use as reference.<br />
Before coming to Tuesday’s precept. Before coming to Tuesday’s precept, you should work through<br />
the Precept Handout and answer the small set of programming questions, called Precept Questions<br />
that are included in the handout. You are allowed to collaborate in only a limited fashion with others. See<br />
below for details. These questions are designed to check whether you understood the materials covered in<br />
the Precept Handout and facilitate problem-solving during precept sessions. You must turn in your<br />
computer code by 10:00 pm the night before precept (according to Submission of Computer Code).<br />
The assessment is graded only on a submitted or not basis, but only “good faith” submissions will count<br />
for full credit.<br />
During precept. The precept will begin by briefly reviewing the handout. Come to precept with<br />
questions if you have any. We will give special attention to issues that seemed problematic for students in<br />
the precept questions. The remainder of the precept time will then be spent working on the new material<br />
and solving Precept Exercises if time permits. Precept Worksheets will be distributed via Blackboard.<br />
You should work through these exercises on your own whether we work on them in the precept or<br />
not. These exercises are a good model for the kinds of problems asked on the Problem Sets and the<br />
Midterm Exam. Solutions to the Precept Exercises will be posted to Blackboard before the following<br />
precept. You will want paper copies of the Precept Worksheets as a reference.<br />
Problem Set/Midterm Exam Format<br />
Each week will end with the posting of either a problem set or the midterm exam. These assignments will<br />
be posted on Thursday afternoons via Blackboard.<br />
Submission. Hard copies of your problem sets must be turned in at the beginning of the Tuesday precept<br />
that they are due. Electronic submission via Blackboard of your computer code must also be done by the<br />
beginning of the Tuesday precept at 9:00am.<br />
Collaboration Policy<br />
The assignments in this course are designated as individual or group assessments. The degree of permissible<br />
collaboration depends on the kind of assignment:<br />
1. Ungraded problems/exercises<br />
2. Precept Questions from Precept Handouts<br />
3. Problem Sets and the Midterm Exam<br />
4. Final Project<br />
Assignment Types<br />
Ungraded problems. For ungraded problems, like those on the Precept Worksheets or old problem<br />
set questions, students are encouraged to interact with each other, the instruction team, and QuantLab tutors<br />
in discussing their approaches and solutions. This includes conceptual discussion and actual computer<br />
code. However, for all other assignments, this degree of collaboration is not appropriate!<br />
3
Precept Questions. Some collaboration on the Precept Questions from Precept Handouts is allowed.<br />
Students may discuss approaches to solving these problems with other students. They may also ask<br />
questions of QuantLab tutors, the instruction team, and on Piazza. However, when it comes time to write<br />
the code that is submitted for precept participation credit, all code must be each student’s own work.<br />
Problem Sets and the Midterm Exam. Problem Sets and the Midterm Exam do not allow<br />
for any collaboration with other students whatsoever. Students may ask clarifying questions regarding<br />
problem set and midterm questions to the instruction team (Jonathan Olmsted, Bella Wang, and <strong>Kosuke</strong><br />
Imai) through Piazza. This allows all students to benefit from clarifications equally. Clarifying questions<br />
about the problem sets may not be asked of QuantLab tutors, however.<br />
Final Project. The Final Project is a group visualization project. Students may fully collaborate within<br />
their assigned groups, but may not discuss their group’s work or another group’s work with other students<br />
in the course.<br />
Plagiarism Policy<br />
Violations of the above collaboration policy will be treated as instances of plagiarism. This course will<br />
follow a modified version of the guidelines used for computer science classes here at <strong>Princeton</strong>. Please<br />
take this guideline seriously. In the past, plagiarism cases typically result in one-year suspension from<br />
<strong>Princeton</strong>.<br />
Programming necessitates that you reach your own understanding of the problem and discover a path<br />
to its solution.<br />
Do not, under any circumstances, copy another person’s code. Incorporating someone<br />
else’s code into your program in any form is a violation of academic regulations. Abetting plagiarism or<br />
unauthorized collaboration by sharing your code is also prohibited. Sharing code in digital form is an<br />
especially egregious violation: do not e-mail your code to anyone.<br />
Novices often have the misconception that copying and mechanically transforming a program (by<br />
rearranging independent code, renaming variables, or similar operations) makes it something different.<br />
Actually, identifying plagiarized source code is easier than you might think. For example, there exists<br />
computer software that can detect plagiarism.<br />
This policy supplements the <strong>University</strong>’s academic regulations, making explicit what constitutes a<br />
violation for this course. <strong>Princeton</strong> Rights, Rules, Responsibilities handbook asserts:<br />
The only adequate defense for a student accused of an academic violation is that the work<br />
in question does not, in fact, constitute a violation. Neither the defense that the student was<br />
ignorant of the regulations concerning academic violations nor the defense that the student was<br />
under pressure at the time the violation was committed is considered an adequate defense.<br />
If you have any questions about these matters, please consult a member of the instruction team.<br />
Textbooks<br />
In addition to the Precept Handout and Precept Worksheets which serve as an introduction to,<br />
discussion of, and reference for the material in this course, there will be assigned readings each week. Some<br />
readings may be distributed (whether as a hard copy or via Blackboard) as appropriate. In the first few<br />
weeks as we learn statistical concepts, readings will come from:<br />
• Freedman, David, Robert Pisani, and Roger Purves. (2007). Statistics. W. W. Norton & Company.<br />
4
This book will be referred to as FPP hereafter.<br />
In addition, the following two books may be helpful references although they are completely optional<br />
for this self-contained course.<br />
• Verzani, John. (2005). Using R for Introductory Statistics. Chapman & Hall.<br />
• Verzani, John. (2011). Getting Started with RStudio. Safari Books Online (Series).<br />
Verzani (2005) is available on reserve at Firestone library: http://goo.gl/tYj5B. Verzani (2011)<br />
is available online through the <strong>Princeton</strong> <strong>University</strong> Library website: http://goo.gl/zKLqzj. You can<br />
access it from any computer on the <strong>Princeton</strong> network.<br />
Finally, at the beginning of the Precept Handouts and Precept Worksheets there will be<br />
references to any corresponding Verzani (2005) sections. In addition, throughout the materials, there will<br />
be pointers to specific sections in Verzani (2011). When you see these Verzani (2011) references, you should<br />
read the corresponding (brief) section online.<br />
Statistical Software<br />
In this course, we use the open-source statistical software R (http://www.r-project.org). Recently, a<br />
New York Times article (“Data Analysts Captivated by R’s Power” January 6, 2009) featured R as,<br />
a popular programming language used by a growing number of data analysts inside corporations<br />
and academia. It is becoming their lingua franca [...] whether being used to set ad prices, find<br />
new drugs more quickly or fine-tune financial models. Companies as diverse as Google, Pfizer,<br />
Merck, Bank of America, the InterContinental Hotels Group and Shell use it. [...] “The great<br />
beauty of R is that you can modify it to do all sorts of things,” said Hal Varian, chief economist<br />
at Google. “And you have a lot of prepackaged stuff that’s already available, so you’re standing<br />
on the shoulders of giants.”<br />
R can be more powerful than other statistical software such as SPSS, STATA and SAS, but it can also<br />
be more difficult to learn. A variety of resources will be made available for POL 245 students in order to<br />
learn R as efficiently as possible.<br />
To help make using R easier, we’ll be using RStudio (http://www.rstudio.com/)—a user-interface<br />
that simplifies many common operations.<br />
Computing Resources<br />
In the event that you need to install R and RStudio on your own personal laptop, this will be covered<br />
in Precept Handout 0 which is due before the first lecture. However, many students may find the<br />
additional computer clusters around campus convenient to use. In these cases, the R and RStudio software<br />
is already installed on the machines in several computer clusters.<br />
• Frist 200-level computer clusters<br />
• Butler D033<br />
Be sure to check the hours of the building’s availability before making plans to use a specific location.<br />
5
Getting Help<br />
POL 245 will be a challenging course and several resources additional are available to facilitate efficient<br />
learning about data analysis and data visualization. We encourage you to take advantage of them whenever<br />
you have questions about the course materials and are struggling with problem sets.<br />
• QuantLab: In addition to the instruction team, there are QuantLab tutors available who can<br />
help you work through the course materials. The tutors are Politics graduate students who have<br />
experience with R and statistics. Unless otherwise noted, QuantLab will be held on Sunday, Monday,<br />
and Wednesday nights from 7:00pm – 10:00pm. Shorter sessions of QuantLab will also be held on<br />
Tuesday and Thursday afternoons from 2:00pm – 3:30pm. Tutoring sessions will be held in Butler<br />
College D026 and D027.<br />
• Office hours: Each of us will hold weekly office hours, starting the first week. In addition to these<br />
office hours, we have an open door policy—you can stop by our Corwin offices at any time without<br />
an appointment and we will answer your questions so long as we are available. You may also e-mail<br />
to set up an appointment with either of us outside of our office hours.<br />
• Discussion board: In addition to office hours and individual appointments, we will be available<br />
online to answer any questions you may have about the course materials and the problem sets. We<br />
use the Piazza discussion forum that will be linked on Blackboard course page or accessible directly<br />
at http://piazza.com. Before posting your question, please review previous posts to make sure that<br />
a similar question has not been answered. In accordance with the collaboration policy described above,<br />
you should not directly post your code for a problem set. You should frame your questions in general<br />
terms rather than trying to have us debug your code directly. You may subscribe to the Discussion<br />
Forum so that you receive your fellow students’ questions and answers to those questions. You<br />
should also feel free to respond to questions that you can answer. Piazza also has a free smartphone<br />
application if you are interested.<br />
Additional Accommodations<br />
Students must register with the Office of Disability Services (ODS) (odsprinceton.edu ; 609-258-8840) for<br />
disability verification and determination of eligibility for reasonable academic accommodations. Requests<br />
for academic accommodations for this course need to be made at the beginning of the semester, or as soon<br />
as possible for newly approved students, and again at least two weeks in advance of any needed accommodations<br />
in order to make arrangements to implement the accommodations. Please make an appointment<br />
to meet with me in order to maintain confidentiality in addressing your needs. No accommodations will<br />
be given without authorization from ODS, or without advance notice.<br />
6
Course Outline<br />
Week 1: July 14 – July 18, Causality<br />
During Week 1, we will learn how to make causal claims from datasets. We learn the distinction between<br />
randomized experiments and observational studies. In addition, we introduce R, programming with data,<br />
and familiarize students with basic skills in data analysis by considering experimental data. These tools<br />
will be applied to topics like evaluating strategies for increasing voter turnout and the effect of class size<br />
on educational achievement.<br />
Note, the first session on Tuesday, July 15 will be a lecture in Sherrerd Hall 101 and not a precept.<br />
The second session on Wednesday, July 16 will be a precept in the Friend Center. The third session on<br />
Thursday, July 17 will be a lecture in Sherrerd Hall. And, for Friday, July 18 we will hold a precept in the<br />
Friend Center. These changes affect only the first week of the course.<br />
Handout 0: posted Monday, Jul. 14; due Tuesday, Jul. 15 (no submission)<br />
Handout 1-A: posted Tuesday, Jul. 15; due Wednesday, Jul. 16<br />
Handout 2-A: posted Friday, Jul. 18<br />
Problem Set 1: posted Thursday, Jul. 17<br />
Required Readings:<br />
FPP, Chapter 1: Controlled Experiments.<br />
FPP, Chapter 2: Observational Studies.<br />
FPP, Chapter 3: The Histogram.<br />
FPP, Chapter 4: The Average and the Standard Deviation.<br />
↩→ only Sections 4.1, 4.2, and 4.3 for this week<br />
Week 2: July 21 – July 25, Clustering<br />
During Week 2, we will further learn methods for describing data and how to identify “clusters” of data.<br />
The main example we discuss is ideological polarization in the United States. In addition, we will look at<br />
the distribution of income, democracy, and natural resource endowments across different countries.<br />
Handout 2-A: due Monday, Jul. 21<br />
Handout 3-A: posted Friday, Jul. 25<br />
Problem Set 1: due Tuesday, Jul. 22<br />
Problem Set 2: posted Thursday, Jul. 24<br />
Required Readings:<br />
FPP, Chapter 4: The Average and the Standard Deviation.<br />
↩→ only Sections 4.4, 4.5, 4.6, and 4.7 for this week<br />
FPP, Chapter 7: Plotting Points and Lines.<br />
Week 3: July 28 – August 1, Prediction and Forecasting<br />
In Week 3, we continue extending general R programming, data-analytic, and data visualization skills.<br />
Forecasting federal and state-level electoral outcomes will be used as a substantive motivation for these<br />
new tools.<br />
7
Handout 3-A: due Monday, Jul. 28<br />
Handout 4-A: posted Friday, Aug. 1<br />
Problem Set 2: due Tuesday, Jul. 29<br />
Midterm Exam: posted Thursday, Jul. 31<br />
Required Readings:<br />
FPP, Chapter 8: Correlation.<br />
FPP, Chapter 9: More about Correlation .<br />
FPP, Chapter 10: Regression.<br />
FPP, Chapter 11: The R.M.S. Error for Regression.<br />
FPP, Chapter 12: The Regression Line.<br />
Week 4: Aug 3 – August 8, Text as Data<br />
For Week 4, we start to cover how text can be analyzed as data. These methods allow researchers to process<br />
large amounts of unstructured text and extract meaningful measures of politically important concepts like<br />
partisanship. We use these methods to analyze the Federalist Papers and the press releases of legislators.<br />
Handout 4-A: due Monday, Aug. 4<br />
Handout 5-A: posted Friday, Aug. 8<br />
Midterm Exam: due Tuesday, Aug. 5<br />
Problem Set 3: posted Thursday, Aug. 7<br />
Week 5: Aug 11 – August 15, Network Data<br />
Week 5 focuses on analyzing data with a network structure. With the rise of social networks, network data<br />
is increasingly common. We use algorithms to describe the networks generated by bill (co-)sponsorship in<br />
Congress and the citation of Supreme Court opinions.<br />
Handout 5-A: due Monday, Aug. 11<br />
Handout 6-A: posted Friday, Aug. 15<br />
Problem Set 3: due Tuesday, Aug. 12<br />
Problem Set 4: posted Thursday, Aug. 14<br />
Week 6: August 18 – August 22, Maps and Spatial Data<br />
During the final week, Week 6, we introduce mapping and visualizing geo-coded data. With this, we will<br />
be able to answer questions about how US voters separate themselves by ideology or to what extent the<br />
country is split into “red” or “blue” states.<br />
Handout 6-A: due Monday, Aug. 18<br />
Problem Set 4: due Tuesday, Aug. 19<br />
Final Project: posted Thursday, Aug. 21; due Wednesday Aug. 27<br />
8
POL 245 Schedule<br />
Due to the condensed nature of the course, the semester and weekly schedules (including assignments) can<br />
seem very full. To help with this, we have a Google Calendar with events, due dates, and sessions. The<br />
URL for this calendar is http://goo.gl/D3pKXL. We also outline a typical week of assignments in POL<br />
245 below:<br />
Sun Mon Tue Wed Thur Fri Sat<br />
7:00pm-<br />
10:00pm<br />
QuantLab<br />
9:00am-9:50am<br />
Lecture<br />
7:00pm-<br />
10:00pm<br />
QuantLab<br />
10:00pm<br />
Submit Handout<br />
code<br />
9:00am<br />
Submit Problem<br />
Set code<br />
9:00am<br />
Submit Problem<br />
Set write-up<br />
9:00am-<br />
10:20am<br />
Precept<br />
2:00pm-<br />
3:30pm<br />
QuantLab<br />
9:00am-9:50am<br />
Lecture<br />
7:00pm-<br />
10:00pm<br />
QuantLab<br />
9:00am-<br />
10:20am<br />
Precept<br />
12:00pm<br />
New Problem Set<br />
posted<br />
2:00pm-<br />
3:30pm<br />
QuantLab<br />
9:00am-<br />
10:20am<br />
Guest Speaker<br />
12:00pm<br />
New Handout<br />
posted<br />
Guest Speakers<br />
For four of the six weeks during the course, we will have guest speakers during our Friday sessions. In<br />
addition to their presentation from 9:00am – 10:20am in the Lewis Library Visualization Lab, they will be<br />
on campus for lunch with a small group of students. Students will sign up for lunch with a specific speaker<br />
at the beginning of the course.<br />
FSI Week Date Guest Affiliation<br />
2 7/25<br />
Archie Tse<br />
Deputy Graphics Director<br />
Telling Stories with Data<br />
New York Times<br />
3 8/1<br />
4 8/8<br />
5 8/15<br />
Eurry Kim<br />
Advertising Researcher<br />
Visualizing the Social TV Networks of<br />
Facebook<br />
Bruce Willsie<br />
President/CEO<br />
Big Data, High Speeds, New Insights—How<br />
High Speed Data Visualization is Changing<br />
Politics<br />
Sondra Scott<br />
Head of Macro and Industry Trends<br />
Energy Fundamentals Analysis—How data<br />
drives insight<br />
Facebook<br />
L2<br />
Wood McKenzie Ltd.<br />
9