MCMC in Python: sampling in parallel with PyMC

Question and answer on Stack Overflow.

Comments Off

Filed under software engineering

IDV in Python: adding text callouts to a scatter plot interactively with mpld3

I’ve been pretty interested in the potential of interactive data visualization recently, especially ever since I saw the reaction that the Global Burden of Disease 2010 visualization tool, GBD Compare, received last year. And one promising technology for making this stuff routine is mpld3, a mashup of the Python plotting library matplotlib and the javascript visualization kernel d3. Have I mentioned this before?

The thing about interactive data visualization is that its not always clear what is useful because it excites my reptile brain, and what is useful for more logical reasons. But I was asking a colleague to add some callouts to a (non-interactive) figure recently when I realized that this is a chance for interactivity to be _obviously_ useful. These finishing touches on a graphic often take me tons of time, and using a command-line plotting program just can’t be the right way to do it. How about an mpld3 plugin that lets me add text callouts interactively? And when I’m done, it can “save” the callouts, by creating the necessary Python script to generate them again? Here it is, in a notebook.

Comments Off

Filed under dataviz

MCMC in Python: Estimating failure rates from observed data

A question and answer on CrossValidated, which make me reflect on the danger of knowing enough statistics to be dangerous.

Comments Off

Filed under statistics

MCMC in Python: How to make a custom sampler in PyMC

The PyMC documentation is a little slim on the topic of defining a custom sampler, and I had to figure it out for some DisMod work over the years. Here is a minimal example of how I did it, in answer to a CrossValidated question.

Comments Off

Filed under MCMC

MCMC in Python: How to set a custom prior with joint distribution on two parameters in PyMC

Question and answer on Stackoverflow. Motivated by question and answer on CrossValidated about modeling incidence rates.

Comments Off

Filed under Uncategorized

Change Seminar: Measuring Mortality in Iraq

I gave a seminar last week for the UW computer scientists interested in doing good with technology. It’s a fun crowd. Here are the slides, requested by a regular attendee who couldn’t be there.

Comments Off

Filed under global health

Fact checking with GBD Compare

I’ve been developing a habit of comparing health statistics I hear in the media with the results in GBD Compare. It is nice when they agree, such as in a recent ScienceMag focus on chronic kidney disease, corroborated here: http://ihmeuw.org/1v7i . It would be even better if the cause was known, and the burden could be removed.

Comments Off

Filed under global health