University of London Statistics Seminar
The seminar meets on Friday 5th February, Friday 26th February and
Friday 19th March at 2.00 pm in room B29, Senate House, Malet Street,
Birkbeck College. This is in the basement of the North Wing of Senate
House, immediately South of Birkbeck College Main Buildings.
Friday 5th February
2.00 pm: Professor U. Statdtmüller, University of Ulm
Is the Regression Function Really Smooth?
3.00 pm: Tea
3.30 pm: Professor A. W. Evans, University College London
Fatal Train Accidents on Britain‘s Main Line Railways
Friday 26th February
2.00 pm: Professor R. v.Sachs, Universite' Catholique de Louvain (UCL)
TBA
3.00 pm: Tea
3.30 pm: Dr. B. Hambly, Hewlett Packard Laboratories, Bristol
Diffusion on fractals and the emission of nitrous oxide
from soil
Friday 19th March
2.00 pm: Dr. S. Nadarajah, University of Plymouth
Some Recent Developments in Extreme Value Theory
3.00 pm: Tea
3.30 pm: TBA
University of London Statistics Seminar
Programme for Friday 5th February
2.00 pm: Professor U. Statdtmüller, University of Ulm
Title: Is the Regression Function Really Smooth?
Abstract: Consider a nonparametric regression model Y(i) = g(i/n) +
e(i), 0 <= i <= n, with independent and centered residuals. We want to
decide from the data whether the underlying regression function is
really smooth or contains jumps. Therefore we assume that g=f+h with a
smooth function f and a step
function h. Analysing squared differences of the data formed with
different span sizes we obtain consistent estimators for the total jump
height. Based on this quantity we can test whether the step function
h is zero or not. If not, we are interested in the number and location
of changepoints in g.
3.00 pm: Tea
3:30 pm: Professor A. W. Evans, University College London
Title: Fatal Train Accidents on Britain‘s Main Line Railways
Abstract: This seminar investigates trends in the frequencies and
numbers of fatalities in fatal train collisions, derailments and buffer
overruns on the national railway system of Great Britain in 1967-1997.
The aims are to estimate the likely scale of future fatalities in train
accidents and to estimate the benefits of possible safety measures.
There were 67 fatal accidents in the 31-year period, with 265
fatalities. On present trends, there are projected to be only about
half as many accidents in the next 30 years as in the last 30, with
about a quarter the number of fatalities. Further safety measures are
possible that would reduce casualties still further, but these sometimes
pose a dilemma in being expensive in relation to the number of
casualties they are likely to save.
The seminar meets in room B29, Senate House, Malet Street, Birkbeck
College
---------------------------------------------------------------
Ruediger Kiesel
Department of Statistics
Birkbeck College
University of London
Malet Street
London WC1E 7HX, England
ph.: 0044- (0)171 - 631 - 6342
fax: 0044 - (0)171- 631 - 6344
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|