Another one for the reading list:
The following new articles have just been published in Population Health Metrics
A novel method for estimating distributions of body mass index
Ng M, Liu P, Thomson B, Murray C
Population Health Metrics 2016, 14 :6 (12 March 2016)
Here is something that Google did not help with as quickly as I would have expected: how do I convert start and stop times into the time between events in seconds (or minutes)?
Or for the busy searcher “how do I convert Pandas Timedelta to seconds”?
The classy answer is:
start_time = df.interviewstarttime.map(pd.Timestamp)
end_time = df.interviewendtime.map(pd.Timestamp)
((end_time-start_time) / pd.Timedelta(minutes=1)).describe()
I found it hidden away here: http://www.datasciencebytes.com/bytes/2015/05/16/pandas-timedelta-histograms-unit-conversion-and-overflow-danger/
The following new article has just been published in Population Health Metrics
Small area synthetic estimates of smoking prevalence during pregnancy in England
Szatkowski L, Fahy S, Coleman T, Taylor J, Twigg L, Moon G, Leonardi-Bee J
Population Health Metrics 2015, 13 :34 (9 December 2015)
I found an even better example of the value of Laplace approximation, and its just a small tweak to the example I did a few weeks ago: http://nbviewer.ipython.org/gist/aflaxman/6d0a9ff2441348f3a130
I admit that I’ve been skeptical of the complete rewrite of PyMC that underlies version 3. It seemed to me motivated by an interest in using unproven new step methods that require knowing the derivative of the posterior distribution. But, it is really coming together, and regardless of whether or not the Hamiltonian Monte Carlo stuff pays off, there are some cool tricks you can do when you can get derivatives without a hassle.
Exhibit 1: A Laplace approximation approach to fitting mixed effect models (as described in http://www.seanet.com/~bradbell/tmb.htm)