If you would
like to be tutored in statistics or probability, please
click here.
If you would
like to find easy-to-read books on statistics, quality control,
or design of experiemnts, please click here.
Do you need the expertise of a statistician in helping to plan your legal strategy, experimental design or quality control strategy? Or possibly you need help in inventory control, worker's compensation liability estimates, environmental law suites, or other problems?
If so, you should contact me, Rick Bilonick, PhD, CQE
(rab@nauticom.net) to
discuss your plans. I am available to provide consulting on a project
by project basis. Statistical analysis can be very profitably
applied to planning your projects -- often providing large savings
in time and money and avoiding legal difficulties. It can also be
usefully applied to data already collected to solve current problems --
both financial and legal.
Here are brief descriptions for a few of the projects I have worked on in the last twenty years:
Provided solid support for challenging shipment descrepancies that cost the producer hundreds of thousands of dollars.
Successfully defended the producer against multi-million dollar claim of a failure to meet contract specifications.
Estimated the likely amount of worker's compensation claims to eventually be paid based on historical claim resolution and outstanding claims.
Conducted numerous mine evaluation studies to determine the feasibility of meeting contract specifications. Applied quality control techniques to maintain the integrity of coal samples used as the basis of payment.
Conducted sampling studies to assess compliance with the contract specifications for delivered coal.
Provided assistance in the design of environmental sampling campaigns for the clean up of industrial sites, particulate material sampling (of metallic ores, liquids, etc.) and the use of geostatistical methods.
Versed in the analysis of mechanical sampling system performance and the use of empirical bias tests of sampling methods and equipment used for industrial and environmental sampling.
Determined the likely amount of savings (several million dollars) due to changes in inventory replenishment strategies before actual implementation based on a stratified random sample from an inventory exceeding 50,000 different products.
Designed countless experiments for coal combustion studies. Designs included factorial, fractional factorial, split-plot, mixture among others.
In various projects, I investigated the spatial distribution of acid rain over the northeastern United States and the distribution of trace element concentrations in coal deposits. Also worked on determining the level of contamination of industrial sites.
Have used spatial statistical methods (geostatistics, semivariograms, kriging) in the construction of risk qualified maps for mineral deposits and environmental contaminants.
I have taught numerous college and company courses covering basic
statistics, experimental design, quality control, sampling, and the use
of statistical software (S-Plus
and SAS). I can tailor a course for
your needs.
I graduated in 1979 with a PhD in Statistics from the University of Pittsburgh and
for the last 18 years have provided (and continue to provide)
statistical consulting services to many departments within CONSOL
Inc. (located in Pittsburgh, Pennsylvania, a subsidiary of CONSOL
Energy Group). I am also a Certified Quality Engineer (CQE).
This certification is provided through the American Society for Quality Control.
I can be reached by e-mail: rab@nauticom.net
or phone: 412.831.4509
or fax: 412.922.9749
or mail:
Dr. Richard A. Bilonick
P. O. Box 113564
Pittsburgh, PA 15241
The graph at the top of this page is a Cumulative Sum quality control
chart. The raw data (shown in a sequential plot below) were grouped in
sets of five for computing the sample means. The original data
consisted a series of 50 zeroes followed by 50 ones which represents a
step change at time t = 51. To this series a random error was added
(normally distributed with a mean of 0 and standard deviation of 1).
The Cumulative Sum chart almost detects the change by t = 60 and
definitely by t = 65. By changing the parameters of the Cumulative Sum
chart, the time to detection could be decreased but at the cost of more
false alarms. Looking at the raw data in the sequential plot one might
be tempted to (incorrectly) assume that the change had occurred at
about t = 38. In this case, the system was still in control at t = 38
but we might have tampered with it causing additional problems.
The graphs were created on a Silicon Graphics Inc. O2 Irix workstation using S-Plus and grabbed and cropped using ImageMagick.
This document was last updated on January 15, 1999.
You are visitor:
.
Copyright © 1999 Richard A. Bilonick (rab@nauticom.net)