Ockham’s Razor: When is the Simpler Theory Better
Elliott Sober
Department of Philosophy, University of Wisconsin-Madison
Wednesday, April 19, 2017
5:30PM in RH300
Abstract: Many scientists believe that the search for simple theories is not optional; rather, it is a requirement of the scientific enterprise. When theories get too complex, scientists reach for Ockham’s razor, the principle of parsimony, to do the trimming. This principle says that a theory that postulates fewer entities, processes, or causes is better than a theory that postulates more, so long as the simpler theory is compatible with what we observe. Ockham’s razor presents a puzzle. It is obvious that simple theories may be beautiful and easy to remember and understand. The hard problem is to explain why the fact that one theory is simpler than another tells you anything about the way the world is. In my lecture, I’ll describe two solutions.
Sponsored by the Vassar College Departments of Philosophy, Psychology, Biology, Mathematics & Statistics, and Science, Technology, and Society.
Lessons from 10+ Years of Chart Watching
Kaiser Fung
Creator of Junk Charts, a leading blog on data visualization
Monday, April 10, 2017
2:30PM in RH310
Abstract: Data visualization has become an essential skill in the Data Revolution that we are living through. It has emerged from a part-time hobby of a few data analysts to a full-time specialization in many businesses. As datasets become more and more complex, visual story-telling gains importance. In all of my work, I take the perspective that charts are primary tools for communications and persuasion.
In this talk, I summarize lessons from 10-plus years of chart watching. There will be plenty of examples of good and bad charts, and lots of to-dos and not-to-dos. A general framework known as Trifecta is introduced, first as a way to evaluate data visualization, and then as a way to enhance data graphics.
Part of the Vassar College Department of Mathematics & Statistics colloquium series.
DataFest @ Vassar
April 7-9, 2017
Registration is now closed. But everyone is welcome to join the opening ceremony 6:00-6:30PM on Friday April 7 and the team presentations and award ceremony 1:00-5:00PM on Sunday April 9, both in Blodgett Auditorium. All are also welcome to stop by Kenyon throughout the weekend to see the student teams at work. For more details, visit http://pages.vassar.edu/datafest/.
The Department of Mathematics & Statistics Colloquium
Overview of campaign data and analytics
Jody Heck Wortman, Duke
Tuesday 3/28/2017, 12:30 @ RH 312
This talk is an overview of campaign data and analytics. The speaker will explore ways to answer the following types of questions, with her experience working as a data analyst at the 2016 Clinton Campaign. For battleground states in a presidential election, how to factor demographics etc. into forming different approaches in different states; how to encourage and measure performance of campaign staff; how to help different departments to make data-motivated decisions, e.g. Where should we send a campaign surrogate? Which voters should we talk to and when? Who of the staff is doing great, and who could use some guidance? How can we get more people more access to convenient voting locations? How many pens, paperclips, and clipboards do we need to order for the last four days of the campaign?
If you want to sign up for an individual meeting with Jody, follow this link.
The Department of Mathematics & Statistics Colloquium
Campaign data and analytics – skills preparation and the importance of efficiency
Jody Heck Wortman, Duke
Tuesday 3/28/2017, 9:30 – 11:00 @ RH 312
The first half of the talk covers some good technical skills to know (e.g. interacting with a database from the user side (VAN); interacting with a database from the back end (SQL); data visualization (Tableau, good old Google Docs); R, Python, Macros) and some good qualitative skills to know (e.g. translating data and analytics into actionable recommendations; hiring and managing people; time management; field organizing) to work in campaign data and analytics.
The second half of the talk focuses on the importance of efficiency in campaign data and analytics. How to take on efficient approaches to questions (e.g. when do you need a rigorous answer, and when will simple be enough?; deciding what’s urgent and important; balancing the answer itself vs. the presentation of it). How to achieve actual efficiency in answering a question (e.g. reproducibility: methods that you can tweak and modify in the future for a quick analysis; keeping things streamlined: i.e., try to do it all in SQL rather than SQL–>Excel–>R–>SQL; efficient code (with comments, on github)).
If you want to sign up for an individual meeting with Jody, follow this link.
The Department of Mathematics & Statistics Colloquium
Boosting Variational Inference
Richard (Fangjian) Guo, MIT
Tuesday 1/31/2017, 12:30 @ RH 312
Modern Bayesian inference typically requires some form of posterior approximation, and mean-field variational inference (MFVI) is an increasingly popular choice due to its speed. But MFVI can be inaccurate in various aspects, including an inability to capture multimodality in the posterior and underestimation of the posterior covariance. These issues arise since MFVI considers approximations to the posterior only in a family of factorized distributions. We instead consider a much more flexible approximating family consisting of all possible finite mixtures of a parametric base distribution (e.g., Gaussian). In order to efficiently find a high-quality posterior approximation within this family, we borrow ideas from gradient boosting and propose boosting variational inference (BVI). BVI iteratively improves the current approximation by mixing it with a new component from the base distribution family. We develop practical algorithms for BVI and demonstrate their performance on both real and simulated data. Joint work with Xiangyu Wang, Kai Fan, Tamara Broderick and David Dunson.
If you want to sign up for an individual meeting with Richard, follow this link.
ASA Sponsored Statistics Social Event
All you ever wanted to know about statistics
Ming-Wen An & Monika Hu
Wednesday 1/25/2017, 2:30-3:30 in Math Lounge (RH 305)
Interested in statistics? Come to our statistics social event on Wednesday 1/25, 2:30-3:30 in Math Lounge (RH 305). We will answer questions about classes, resources for finding summer internships/research opportunities, upcoming talks this semester, DataFest 2017 @ Vassar, American Statistical Association membership and student chapter, Joint Statistical Meetings 2017 in Baltimore etc. Tasty snacks and beverages will be provided. We look forward to welcoming everyone back to campus for the new semester!
Save the date: The Department of Mathematics & Statistics Henry Seely White Lecture Series with Dr. Susan Murphy, University of Michigan: February 18-19, 2016. Recent Amstat News article about Susan Murphy.
Electronic Undergraduate Statistics Research Conference: Friday October 2, 2015. Click here for more information about a free conference with exciting topics about statistics research. Presentations by two keynote speakers as well as undergraduate students.
Fall 2012 Workshop on R. On Saturday September 8, 2012, please join us for a workshop on R led by Jay Emerson, Yale University. We will follow a casual schedule, with a morning session focusing on the basic language syntax and building a foundation with simple examples and an afternoon session focusing on more involved examples, hopefully contributed by workshop participants.
R is an open-source statistical software that has been used by statisticians and other scientists for about 15 years and has gained widespread appeal (see NY Times article).
Please note, this event is open to interested Vassar faculty. An e-mail with more details was sent to all faculty. Interested students should e-mail mian@vassar.edu.
Save the Date: March 23-24, 2012. We are pleased to be hosting Professor John Kruschke (Indiana University; personal website) at Vassar College on March 23-24. Professor Kruschke is a passionate advocate of the exciting but as yet relatively little-known Bayesian approach to analyzing data and he will give an engaging and accessible presentation of its basic elements and potential advantages over traditional statistical methods. His visit will include a public talk and a workshop:
Friday March 23–
Public Talk: “Doing Bayesian Data Analysis”
[4pm, Rockefeller Hall 300; with light refreshments at 3:45 outside of Rockefeller Hall 300]
Saturday March 24–
Hands-on Workshop for those interested in learning about the Bayesian approach in greater depth
[9am-12pm, SciVis Lab, Mudd Hall 3rd Floor]
–Please RSVP to mian@vassar.edu with subject line: “RSVP Kruschke workshop” so that we can plan resources accordingly and communicate workshop-specific details.
*For more details on the content of the talk or the workshop, please visit this page.
*This event is sponsored by the Vassar College Dean of Faculty; the Departments of Biology, Economics, Mathematics, and Psychology; the Cognitive Science Program; the Center for Science and Quantitative Reasoning (CSQR); and the Center for Collaborative Approaches to Science (CCAS/HHMI).