At SciPy 2015 I’ll present a short talk on sound processing in Python, based on my book-in-progress, Think DSP, and the class I helped teach this spring. I’ve been preparing the talk this week; it is almost ready to go! The current draft of the slides is here, and the accompanying notebook is here.
Here’s my favorite figure from the talk, the spectrogram of a sawtooth chirp (notice the aliasing bouncing off the Nyquist roof):
To hear what it sounds like, you can load the notebook and play it. The talk is July 8 at 2pm.
Last week I was in São Paulo, Brazil to visit Insper, a small private college in the process of starting a new engineering program. The first semester of the program, with the first group of 90 students, is wrapping up this week.
The engineering faculty at Insper have been working with Olin for several years. Their curriculum is similar to Olin’s in some ways, but of course they have designed it for their students and environment. One awesome thing: all students learn Python in the first semester, as I discovered when I visited their library:
And they all take a version of Modeling and Simulation, a class I helped develop at Olin, along with John Geddes and Mark Somerville. They have an excellent FabLab, which was hopping during the last week of classes:
Visting São Paulo was exciting for me — it was my first time in Brazil, and my first time in South America. I used Duolingo to learn some Portuguese, which turned out to be more useful than I expected. I didn’t get much past “Bom dia” and “Obrigado”, but I found that since I know some Latin, Spanish, and French, I could read a lot of Portuguese, especially more technical material, which uses a lot of cognates. I might even have created a new word: talking about Rube Goldberg contraptions, I suggested “contrapção”. That’s a word now.
The Insper faculty are working hard and creating an amazing new program. And the students are great. I sat in on their intro Python class and had a chance to work with a few student teams. They came up with some great projects, some of them very ambitious for an intro class, including one team using OpenCV for facial recognition.
But I am happy to be home, at least for a couple of weeks. Next trip: Austin TX for SciPy!
In the last few weeks I’ve been to the Netherlands and Chicago, and tomorrow I am off to Brazil. But I spent most of this week in my office, enjoying the luxury of working on just one thing: a paper about survival analysis and marriage patterns in the U.S. I just submitted it for review, so watch this space for more.
Here’s a picture of me outside my office, featuring covers from translations of my books:
Also this week I got a pleasant surprise in the mail, a signed copy of a new book by an old friend, Dror Feitelson: Workload Characterization for Computer Systems Performance Evaluation. Here it is in its new home:
It looks like a valuable collection of topics that don’t usually appear together, but should. It’s published by Cambridge University Press, and Dror also makes the PDF version available for free download. A few parts of the book go all the way back to a paper Dror and I worked on together in 1999, “The Elusive Goal of Workload Characterization,” but it has come a long way since then!
My SciPy strategy has gone horribly wrong. I submitted two talks and a tutorial with the expectation that 1.0 of my proposals would be accepted, on average. But the vagaries of the binomial distribution bit me: all three were accepted. So I am scrambling to get ready.
My tutorial on statistical inference is ready to go. As it turns out, Chris Fonnesbeck proposed a similar tutorial at a more advanced level, so we re-titled our tutorials as Computational Statistics I and Computational Statistics II. We are scheduled back to back, and each of us is planning to help out during the other’s session. Both tutorials are now full!
One of my talks is in the Computational Social Science thread, where I will present “Will Millennials Ever Get Married?” I’ll present results from applying survival analysis to data from the National Survey of Family Growth (NSFG). A basic version of the analysis appears in Chapter 13 of Think Stats, 2nd edition. What I am working on now is a more careful analysis using data from earlier and later cycles of the NSFG.
I am close to finishing off the results I want to present. Here’s a screenshot of the current status:
My second talk is on Digital Signal Processing in Python. This is the only talk I have not presented before, but it is based on Think DSP and the class I taught in the spring. I have tons of material; the hard part will be selecting elements that make a complete and coherent talk.
This week I am one of several Olin professors helping out with the Olin Collaboratory Summer Institute. This year we have 55 participants, including faculty from universities around the country (like Oklahoma and Vermont) and the world (like Saudia Arabia, India, and Australia).
During the first half of the week, we lead workshops where participants learn tools for curriculum design. On Monday Mark Somerville and I lead a workshop on understanding students and designing courses and programs that serve their needs. Here’s a picture of me saying something apparently important:
Olin Summer Institute
During the second half of the week participants work on projects they choose, often redesigning a class or creating a new curriculum. More information about the Summer Institute is here.
This weekend I was at the Open Data Science Conference, here in Boston, for a book signing (with thanks to the nice people at O’Reilly Media) and to present “Learning to Love Bayesian Statistics”. Here’s an audience shot of the presentation (with thanks to @kjchoi101):
My slides are at http://tinyurl.com/lovebayes. Video should be available soon — I just hope they edit the 5-10 minutes I spent getting wireless to work. Next time someone suggests I should download my slides ahead of time, I will listen!