Statistical Computing
Computing at Denver: An Overview
Wolfgang Jank, Program Chair
T
he Section on Statistical Computing because there is no clearly defined common the problem is that practically no useful
has lined up an exciting program subject matter and because there is no characterization of the global maximum is
for JSM 2008. Topics fill the spec- widely used text that prescribes the content. available. This session will present different
trum, from curriculum design to method- This session will present several approaches solutions to global optimization problems
ological development to analysis of real- to designing a broad and sufficient course in the context of the EM algorithm.
world problems. Each session varies in on modern statistical computing. The second methodological session con-
format and includes invited speakers and Two invited sessions deal with the devel- siders advances in functional data analysis
discussants from around the world, aca- opment of statistical methodology. The first (FDA). The technological advancements
demia, research labs, and industry. The session is on global maximization in EM- in measurement, collection, and storage of
common theme among all is the computa- type algorithms. In likelihood-based mod- data have led to more complex data struc-
tional aspect, which could be in the form of eling, finding the parameter estimates that tures. Examples include measurements of
teaching statistical computing to students, correspond to the global maximum of the individuals’ behavior over time, digitized
developing new computational algorithms, likelihood function is a challenging prob- two- or three-dimensional images of the
or using computational methods to analyze lem. This problem is especially serious in brain, and recordings of three- or even four-
challenging and large-scale data. Particular statistical models with missing data, where dimensional movements of objects travel-
thanks go to Deepak Agarwal, David popular EM-type algorithms are used to ing through space and time. Such data,
Banks, Giles Hooker, Ravi Vardhan, and obtain parameter estimates. Despite their although recorded in a discrete fashion, are
Chris Volinsky for putting together such an hallmark stability and monotone conver- usually thought of as continuous objects
exciting program. gence, EM-type algorithms (i.e., the EM represented by functional relationships.
Our first invited session is about algorithm and its variants, such as MM This gives rise to FDA, where the center of
designing courses on statistical computing. algorithms) frequently converge to local, interest is a set of curves, shapes, objects,
By now, most statistics departments have suboptimal solutions, especially when the or a set of functional observations. This is
recognized the need to make computation dimension of the parameter’s space and/or in contrast to classical statistics, where the
a focus of their graduate and undergraduate the size of the data are large. This is a par- interest centers on a set of data vectors.
education. This has been done in different ticular concern in finite mixtures and latent FDA has experienced rapid growth over the
ways, but the most common is a specifically variable models, for example. The problem past few years, both in the range of appli-
designed course taught in the first or second has largely been ignored in statistics, and cations for its techniques and in the devel-
semester of a student’s career. The contents of only unprincipled approaches have been opment of theory for statistical inference.
such a course are not well established, in part used (e.g., random multi-starts). Part of This session will bring together leading
38 AMSTAT NEWS JANUARY 2008
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84