Readme file
SERIES B
 Statistical Methodology
One-step
local quasi-likelihood estimation,
J. Fan and J. Chen
Journal of the Royal Statistical Society, Series B, Volume 61, (1999), Part 4,
927-943
DESCRIPTION
OF THE DATA:
This is an enviromental dataset, consisting of
daily measurements of pollutants and other environmental factors in Hong Kong between
January 1, 1994 and December 31, 1995. The association between levels of pollutants
and number of daily hospital admissions for circulation and respiration problems is of
particular interest.
In this paper, we apply our methods for the
Binary regression and Poisson regression to investigate respectively how the
probability of high level Sulphur Dioxde SO2 (with values > 20 ug/m^3) is
associated with level of polutant Nitrogen No2 (in ug/m^3) and how the number of hospital
admissions is associated with the level of NO2. This data set is in files.txt.
COMPUTER PROGRAMS:
There are two computer programs in this paper,
known as the Binary program ( binary.c ) and the Poisson program ( poisson.c)
respectively.
One can use the program to implement one-step
quasi-likelihood estimation for the Binary regression and the Poisson regression. It may
be used to perform real data analyses and simulation experiments. The programs
automatically choose optimal bandwidths for the one-step estimator, the weight least-
squares estimator and the maximum likelihood estimator. The comparisons of three methods
also are given. The programs themselves are in files.txt. We will introduce how to use the
programs as follows:
(1) Run a "c program"
For example:
%cc binary.c -lm (Enter)
% /a.out (Enter)
One-step estimation for logistic regression
Which Example (0 -- 3) ?
Note: Example 0 is for real data
(2) Choose to perform real data analysis or the
simulations (Example=0 means real data analysis; Example=1,2,3 means simulation). Then
input the parameter values.
n=sample size, Nsim=number of the simulation,
ngrid=number of grid point,
xgridmin=minimum value of the grid point,
xgridmax=maximum value of the grid point.
(3) Input data filename and output filename.
For example:
Which Example (0 -- 3) ?
Example 0 is real data example: 0
Enter sample size n, ngrid, xgridmin, xgridmax:
Enter input filename:
Enter OUTPUT filename:
or
Which Example (0 -- 3) ?
Example 0 is real data example: 1
Enter n, Nsim:
Enter OUTPUT filename:
Whether estimatng the optimal bandwidth from
DATA?
(1=YES): 1 (1 means that the optimal handwidth is
known
and input its value)
Enter optimal handwidth:
One-step estimation for logistic regression
Example 1:
Contact Addresses:
Prof. Jianqing Fan
Department of Statistics
University of North Carolina at Chapel Hill
Chapel Hill, NC 27599-3260
E-mail: jfan@stat.ucla.edu
http://www.stat.unc.edu/faculty/fan.html
or http://www.stat.ucla.edu/~fan
Jianwei Chen
Department of Statistics
Chinese University of Hong Kong
Shatin, N.T. Hong Kong
E-mail: jwchen@hp735.sta.cuhk.edu.hk
|
