Three-Toed Sloth
http://bactra.org/weblog/
Slow Takes from the Canopy (My Very Own Internet Tradition)en36-835, Paper of the Week
http://bactra.org/weblog/682.html
<P>As threatened, I'll post links to the paper being discussed each week in the
<a href="673.html">statistical modeling seminar</a>. This will happen
after the discussion, and my own brief comments here will also not be shared
with the students beforehand. <a href="682.rss">This</a> should be an RSS feed
for this page.
<ol>
<li>Leo Breiman, "Statistical Modeling: The Two Cultures",
<a href="http://projecteuclid.org/euclid.ss/1009213726"><cite>Statistical
Science</cite> <strong>16</strong> (2001): 199--231</a>
<br><em>Comment</em>: I remember being very excited by this paper when it came
out. The students were less taken with it — "<em>Of course</em> you use
cross-validation to check predictive performance, why does he feel like he has
to say that?" In retrospect, I would say that what Breiman calls "data models"
are very rarely serious scientific models of the data-generating mechanism, but
more "algorithmic models" of a pre-computer age...
<li>Sarat C. Dass and Mingfei Li, "Hierarchical mixture models for assessing fingerprint individuality", <a href="http://projecteuclid.org/euclid.aoas/1267453947"><cite>Annals of Applied Statistics</cite> <strong>3</strong> (2009): 1448--1466</a>
<br><em>Comment</em>: This is interesting, but the big problem is that they did
absolutely nothing to convince me that their model works.
(<a href="http://arxiv.org/abs/1006.3868">Cf.</a>) Consequently, why should I
think that their estimates of false-identification probabilities are even
roughly right? (Also, why not model a spatial point process as a spatial point
process?)
</ol>
<P><span class="blognotes">
<a href="http://bactra.org/weblog/cat_corrupting_the_young.html">Corrupting the Young</a>;
<a href="http://bactra.org/weblog/cat_enigmas_of_chance.html">Enigmas of Chance</a>
</span>