Newsgroup sci.stat.math 14406

Directory

CESA'98 -- Call for Papers for the Invited Session on
            +------------------------------------------------+
            |        Intelligent Prognostic Methods in       |
            |    Medical Diagnosis and Treatment Planning    |
            +------------------------------------------------+
                                  during
            Computational Engineering in Systems Applications
               CESA '98 - IEEE/SMC-IMACS, 1 - 4 April 1998
                 (http://lails1.ec-lille.fr/~cesa98/)
                        Nabeul-Hammamet, Tunisia
  (WWW version of the CfP: http://www.cs.ruu.nl/~lucas/ipm-cesa98.html)
Important Dates:
---------------
Submission deadline:            12 September 1997
Notification of acceptance:     29 September 1997
Conference:                     1 - 4 April  1998
Computational prognostic models are increasingly used in medicine to
predict the natural course of disease, or the expected outcome after
treatment.  Prognosis forms an integral part of systems for treatment
selection and treatment planning. Furthermore, prognostic models may
play an important role in guiding diagnostic problem solving, e.g. by
only requesting information concerning tests, of which the outcome
affects knowledge of the prognosis.
In recent years several methods and techniques from the fields of
artificial intelligence, decision theory and statistics have been
introduced into models of the medical management of patients
(diagnosis, treatment, follow-up); in some of these models, assessment
of the expected prognosis constitutes an integral part.  Typically,
recent prognostic methods rely on explicit (patho)physiological
models, which may be combined with traditional models of life
expectancy. Examples of such domain models are causal disease models,
and physiological models of regulatory mechanisms in the human
body. Such model-based approaches have the potential to facilitate the
development of knowledge-based systems, because the medical domain
models can be (partially) obtained from the medical literature.
Various methods have been suggested for the representations of such
domain models ranging from quantitative and probabilistic approaches
to symbolic and qualitative ones. Semantic concepts such as time, e.g.
for modelling the progressive changes of regulatory mechanisms, have
formed an important and challenging modelling issue. Moreover,
automatic learning techniques of such models have been proposed. When
model construction is hard, less explicit domain models have been
studied such as the use of case-based representations and its
combination with more explicit domain models.
This invited session aims at bringing together various theoretical and
practical approaches to computational prognosis, possibly in the
medical setting of diagnosis and therapy planning, that comprise the
state of the art in this field. Papers are sought that describe
medical prognosis applications using methods and techniques from
artificial intelligence, decision theory, and statistics as well as
papers proposing theoretical foundations of such methods.
The best papers will entitle their authors to an invitation to submit
an extended paper for a special issue on "Intelligent Prognostic
Methods in Medicine" of the journal of "Artificial Intelligence in
Medicine".
TOPICS OF INTEREST
Papers are sought on topics including, but not limited to:
o Modelling and Reasoning:
  - ontologies for the specification of medical domain models
  - the specification of prognostic models, possibly as part of 
    diagnostic or therapy-planning applications 
  - representation and reasoning about (multiple) model types such as
    empirical, anatomical and (patho)physiological ones 
  - representation of and reasoning with time
  - qualitative representation and reasoning
  - (dynamic) probabilistic networks
  - decision analytical modelling
  - function-based representation and reasoning
  - case-based representation and reasoning
o Knowledge Acquisition:
  - acquisition of the medical prognostic models
  - automated learning of domain or task models using machine learning 
    techniques 
o Use and Reuse of Prognostic Models in:
  - development of medical guidelines and protocols
  - medical diagnostic systems
  - treatment selection and planning systems
o Formalisation:
  - use of logical, set-theoretical or probabilistic methods to formalise 
    various aspects of prognosis and therapy planning 
o Medical Applications:
  - clinical context of actual prognostic models
  - role of prognostic models in diagnosis or treatment planning of 
    a specific disease 
  - evaluation of prognostic models
TIME SCHEDULE AND PUBLICATION
Submission deadline:            12 September 1997
Notification of acceptance:     29 September 1997
Camera ready version:           15 January 1998
Conference:                     1 - 4 April  1998
Each submission will be refereed by at least two members of the
programme committee. Accepted papers will be published in the
proceedings of CESA'98 belonging to the Symposium on Signal Processing
and Cybernetics.  The special issue on "Intelligent Prognostic Methods
in Medicine" of the journal of "Artificial Intelligence in Medicine"
will be published in mid 1999.
INSTRUCTIONS TO AUTHORS
Contributions, not exceeding 8-10 pages (about 5000 words) are to be
addressed to the first co-chair and should be written in English with
an abstract and a list of keywords. Electronic submissions by e-mail
are encouraged (either postscript files or plain text). Alternatively,
3 paper copies may be submitted. Camera-ready manuscripts should not
exceed 6 pages prepared in accordance with CESA's "Guide for Authors"
(a LaTeX style file is available).
Authors planning to submit a paper to the invited session, and
indirectly to the special issue of the journal "Artificial
Intelligence in Medicine" are requested to contact one of the
co-chairs as soon as possible.
SESSION ORGANIZATION
Co-Chairs: 
Ameen Abu-Hanna, University of Amsterdam, The Netherlands
Peter Lucas, Utrecht University, The Netherlands
PROGRAMME COMMITTEE
The programme committee currently consists of:
A. Abu-Hanna, The Netherlands           
L. Console, Italy
G.F. Cooper, USA
J. Fox, UK
P. Hammond, UK
R. Haux, Germany
E. Keravnou, Cyprus
N. Lavrac, Slovenia
P.J.F. Lucas, The Netherlands
M. Musen, USA
M. Ramoni, UK
M. Stefanelli, Italy
J. Wyatt, UK
For more information about the invited session, the conference or the
special issue in the journal of "Artificial Intelligence in Medicine"
please contact one of the co-chairs.
Ameen Abu-Hanna                      Peter Lucas
Dept. of Medical Informatics         Dept. of Computer Science
Academic Medical Center              Utrecht University
University of Amsterdam              Padualaan 14
Meibergdreef 15                      3584 CH Utrecht
1105 AZ Amsterdam                    The Netherlands
The Netherlands
E-mail: A.Abu-Hanna@amc.uva.nl       E-mail: lucas@cs.ruu.nl
Telephone: +31 20-5664511            Telephone: +31 30 2534094
Fax      : +31 20-6912432            Fax:       +31 30 2513791
--
Peter Lucas
Dept. of Computer Science, Utrecht University
Padualaan 14, 3584 CH Utrecht, The Netherlands
Tel: + 31 30 2534094; E-mail: lucas@cs.ruu.nl

In article <866539660.29224@dejanews.com>, spurling@sonoma.edu writes:
>In article <19970616173201.NAA18338@ladder02.news.aol.com>,
>  dnordlund@aol.com (DNordlund) wrote:
>> One thing that I find interesting about the standard Monte Hall problem
is
>> that the naive person who thinks that the probability of winning
changes
>> from 1/3 to 1/2 (not 2/3) after a door is opened can actually obtain
>> "evidence" that this is true.  The person reasons that the probability
of
>> winning is 1/2, so it doesn't matter whether or not a switch is made. 
So
>> by some 'random' process the person switches 50% of the time.
>>
>> The person wins with probability 1/2!  QED.  :)
>>
>> In fact, unless one switches all the time, it will take many trials to
>> demonstrate empirically that the probability is not 1/2.  By switching
3/4
>> of the time the probability of winning only increases to approximately
>> 0.58.
>>
>Dan:
>
>Let me think about that one a bit.  In the meantime, let me continue on
>with this example.  In problem 1 there was no reduction in choice and
>there was no relative advantage to switching over staying.  In the second
>problem because of the reduction in choice, switching was advantageous to
>staying.  Let me ask if this advantage is conditional on your knowing
>that a reduction in choices has been made? By way of clarification, let
>me change the problem to this.	
>Instead of asking you to choose a number
>other than 2, I merely give you a die loaded in such a way that the
>number 2 does not come up (a number that does not produce a win anyway)
>and ask if you want to roll again..  The other five numbers are now
>equiprobable.  Is there still a relative advantage to rolling again
>rather than staying with your original roll even though you don't know
>that there has been a reduction in the number of incorrect choices?  
Here the problem has become what I originally thought problem 2 was.  My
choice is to roll the die again or not.  The probability of getting a
match becomes 1/5 by rolling with the loaded die.  So, it would be to my
advantage to roll again.  However, I don't know the die is loaded, there
is no reason for me to think that the probabilities have changed,
therefore it doesn't matter whether I roll or stay.
>And if the answer is yes, then in what situation if any would staying be
>advantageous to switching?  That is if switching is always at least as
>good as staying, then why wouldn't somebody always switch?
>
In the classic Monte Hall problem, one SHOULD always switch; it is never
advantageous to stay.  This presumes that Monte always opens an empty
unchosen door.
In the Monte Hall type problem, the advantage in switching does not come
just from a reduction in the number of choices, it depends on how the
reduction occurs.  If Monte Hall 'randomly' chooses to open one of the
non-selected doors, 1/3 of the time he would open a door with a prize
behind it, and you lose.  On the remaining 2/3 of the trials you would win
1/2 of the time. Therefore overall, you win 2/3 * 1/2  = 1/3 of the time.
If you always switch you will still only win 1/3 of the time. Switching
makes no difference if Monte opens a door randomly.
Dan

StatSoft, Inc., a leading manufacturer of statistics software 
(STATISTICA) is seeking qualified applicants to fill several positions 
in our US and Japan offices, all available immediately.  Opportunities 
exist in technical support, customer sales, assisting in program design, 
software testing, developing program documentation, and teaching courses 
and seminars.
Applicants are sought from any field of science (e.g., Statistics, 
Mathematics, Computer Science, Psychology, Biology) for the following 
positions:
Application Developer	
Applicant must have a degree in computer science or related field, 
experience with C, FORTRAN, and C++.  The ideal applicant will be 
familiar with programming in the Windows environment, easy to work with, 
and capable of quickly learning new concepts.
Senior Statistical Developer	
Applicant must have a degree in computer science or related field and a 
M.S. or Ph.D. in statistics or related field.  Applicant much have 
experience in numerical analysis programming.  The ideal applicant will 
also be familiar with programming in the Windows environment.
Technician	
Applicant must have a minimum of 9 semester hours of statistics (or 
equivalent), extensive microcomputer experience, and experience with at 
least one commercial statistical analysis package.  Excellent written 
and oral communication skills are required.  The ideal applicant will be 
familiar with the Windows environment and application software.
Consultant	
Applicant must have a M.S. or Ph.D. in statistics or related field, 
extensive microcomputer experience, and experience with data analysis 
software.  Excellent written and oral communication skills are required, 
as well as experience with data analysis in any area of statistics.  The 
ideal applicant with be familiar with the Windows environment 
application software.  Programming experience is a plus.
Please submit your application by mail, fax (attention: Personnel), or 
e-mail: jobs@statsoft.com
StatSoft
Attn: Personnel
2300 E. 14th Street
Tulsa, OK 74104
phone: 918-749-1119
fax: 918-749-2217
StatSoft, Inc. is an Equal Opportunity Employer.
-- 
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
*  Win Noren                    *   phone : 918-749-1119        *
*  Technical Services Manager   *   fax   : 918-749-2217        *
*  StatSoft, Inc.               *                               *
*                               *   2300 E. 14th Street         *
*  e-mail: win@statsoft.com     *   Tulsa, OK 74104             *
*  www: http://www.statsoft.com *                               *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

who could please give me a quick hint which procedure is used to compute
the following:
I have got a series of measures varying about a certain value.
say: 48,49,43,51,47, 49,48,...etc.
now there are a few numbers completely out of range due to systematic
errors,
say: 1,2 and 178,211 (let me call them "tear out's" :-) )
If I plot the histogram, I can see those concentratet in intervals at both
ends of the range and the ones I want to extract are well distributet in
the middle of it.
How can I find (compute) the intervals to delete and the intervals to leave
in order to proceed with statistics on them?
any (tiny) hint is very much appreciated (just the name of any test is
fine),
yours
Joern.
-----------------------------------
Joern.Apel@post.rwth-aachen.de
Helmholtz-Institute f. Biomechanics
Tel.: ++49-241-80-89416
Fax.: ++49-241-8888442
-----------------------------------

  Hrubin said:
 >According to present physical theory, nature is even worse than random<
Such an interesting thought.  I think this allows me to ask the question,
"Is life fair?"
I am very interested in any response.
     Mod Tollens  ......................  in the meantime, I am going to
the Poker discussion to apologize for my error.

---------------------------------------------------------------------
        ANNOUNCEMENT and SECOND CALL FOR PAPERS - IMVIP '97
     IRISH MACHINE VISION & IMAGE PROCESSING CONFERENCE 1997
                    September 10-13, 1997
                     University of Ulster
                        Magee College
                       in tandem with
   NINTH IRELAND CONFERENCE ON ARTIFICIAL INTELLIGENCE - AI-97
---------------------------------------------------------------------
GENERAL INFORMATION
The Irish Machine Vision & Image Processing Conference is a forum for
presentation of original research, both basic and applied, and exchange
of ideas in a broad area of image processing, image communication,
pattern recognition and machine vision. 
The conference has developed from a highly successful Machine Vision
and Image Processing Colloquium staged at the Queen's University
Belfast in March 1996. The technical programme will comprise tutorials,
plenary lectures, an exhibition, and contributed papers, which will be
presented in either lecture and poster format.
IMVIP '97 will take place at Magee College, Derry, Northern Ireland,
from Wednesday 10th to Saturday 13th September 1997. (Please note
changed
dates, compared to an earlier Call.)  A conference dinner will be held
on 
the Friday evening. Persons requiring accommodation may avail of the 
University student residences for bed and breakfast. A list of local 
hotels and B&Bs; is also available at the conference Web address.
Conference URL:     http://www.infm.ulst.ac.uk/research/imvip97
AI-97 URL:          http://www.infm.ulst.ac.uk/research/ai97
Conference e-mail:  imvip97@ulst.ac.uk (or addresses below)
Deadline for submission of abstracts: Sunday 29 June 1997.
We would be grateful if you would copy this notice to colleagues
who may be interested in IMVIP '97.
With support from: 
ECVnet - European Computer Vision Network of Excellence.
IAPR - International Association for Pattern Recognition.
IRTU - Industrial Research and Technology Unit of Northern Ireland.
OESI - Optical Engineering Society of Ireland, Cumann Innealto/ireacht
       Optu/la (Irish chapter of SPIE).
----------------------------------------------------------------------
TOPICS and THEMES
Topics and themes include but are not limited to:
- General techniques and algorithms: image filtering, enhancement;
  pattern recognition, neural networks, fuzzy sets.
- Industrial inspection for manufacturing and processing.
- Biomedical image processing.
- Document image processing.
- Remote sensing and space applications.
- Multimedia, including WWW-based image processing.
- Image analysis in security and surveillance, transport.
- Digital images in video, television, telephony.
- Systems: software, hardware, architectures.
Papers in broadly associated areas are encouraged. 
--------------------------------------------------------------------
SUBMISSION OF PAPERS
Prospective authors are invited propose papers on any of the topics
mentioned. Send FOUR (hardcopy) copies of a ONE page abstract, together
with completed cover form (see below) to:
IMVIP '97 Secretariat,
School of Information & Software Engineeering,
University of Ulster, Magee College, 
Londonderry, BT48 7JL.
Accepted papers will be published in the Proceedings of IMVIP '97,
which will be available at the time of the conference; an eight-page
limit will be imposed on the final papers.
---------------------------------------------------------------------
SCHEDULE
29 June 1997      Deadline (extended) for submission of abstracts.
13 July 1997      Notification of acceptance.
17 August 1997    Submission of camera-ready paper.
10-13 September   Conference and Tutorials.
--------------------------------------------------------------------
TECHNICAL PROGRAMME COMMITTEE
Honorary Chairs: Prof F Monds (UU), Prof JG Byrne (TCD)
Dr E Ambikairajah (Athlone RTC), Prof DA Bell (UU), Dr N Black (UUJ),
Dr A Bouridane (QUB), Mr J Campbell (UUM), Mr T Carew (TSI), Prof D
Crookes (QUB), Dr K Dawson-Howe (TCD), Prof A Hashim (UUM), Prof J
Haslett (TCD), Dr P Horan (DIT), Prof C Hussey (UL), Dr J
Keating (Maynooth), Dr J Kennedy (CAPTEC), Mr G Lacy (TCD), Prof P
McKevitt (Aalborg), Dr N McMillan (Carlow RTC), Dr P Morrow (UUC), Dr N
Murphy (DCU), Prof F Murtagh (UUM), Mr F Shevlin (TCD), Prof F
O'Sullivan (UCC), Prof D Vernon (Maynooth), Dr P Whelan (DCU).
--------------------------------------------------------------------
LOCAL ORGANISING COMMITTEE
Mr J Campbell (UUM), Ms J Farren (UUM), Ms C McNutt (UUM), Dr P Morrow 
(UUC), Prof F Murtagh (UUM).
-------------------------------------------------------------------
CONTACT
Ms C McNutt, 
Faculty of Informatics, University of Ulster, Magee College,
Londonderry BT48 7JL, Northern Ireland.
Fax +44 1504 370040 (from Republic of Ireland: 080 1504 370040)
Tel +44 1504 375408
           / 375367 (answering machine on 375367) - Jon Campbell.
E-mail:         imvip97@ulst.ac.uk
Conference URL: http://www.infm.ulst.ac.uk/research/imvip97
---------------------------------------------------------------------
KEYNOTE PRESENTATIONS
- Prof James L Crowley, INPG, Grenoble, France. Coordinator, European
  Computer Vision Network of Excellence:
           "Computer Vision for Man-Machine Interaction".
- Prof Anil Jain, Department of Computer Science, Michigan State
  University:
          "Image and Video Databases".
- Dr Jean-Christophe Olivo, European Molecular Biology Laboratory, Cell
     Biophysics Programme, Heidelberg: 
          "Image Processing Applied to Biological Images".
AND AI-97 KEYNOTE PRESENTATIONS
- Prof John McCarthy, Department of Computer Science, Stanford
University, 
  Stanford, CA, US. 
- Prof Dr Walther von Hahn, Department of Computer Science, University 
  of Hamburg, Germany. 
- Prof Naoyuki Okada, Department of Computer Science, Kyushu Institute 
  of Technology, Iizuka, Japan. 
---------------------------------------------------------------------
AI-97/IMVIP-97 PLENARY LIVE FEED
It is intended  that the main plenary  sessions at AI/IMVIP  go out on
streaming  video and audio, stored   and live with  the possibility of
phone-in questions (organisation: Ted Leath, Magee College).
SOCIAL PROGRAMME
There will be a reception on Thursday 11 September, a conference banquet 
on Friday 12 September, and a trip to the Giant's Causeway and Bushmills
Distilery on Sunday 14 September. 
CONFERENCE VENUE 
Set beside the meandering River Foyle where it becomes Lough Foyle, 
Derry or Londonderry (and some other names besides) has a rare scenic
beauty.  It is rich in history, encompassing monastic settlement and 
fully extant city walls, the great seige of the late 17th century, and
much more.  A visit to the renowned Tower Museum is more than 
rewarding, as is a visit to the rugged mountains and sea cliffs in the
close hinterland of Donegal.  It is a northern European city of 100,000, 
almost on the border between the Republic of Ireland and Northern 
Ireland.  It has wide renown for its writers and musicians.
Informatics at Magee College is building up a strong programme in the 
areas of computational intelligence, intelligent multimedia, and 
distributed object computing.  See www.infm.ulst.ac.uk/research
VENUE
The venue for registration, posters and exhibits, and for all conference 
events, will be MG 220 and MG 229 in the MG Building.  Magee College 
itself is a short walk from the city centre. 
TRANSPORT 
>From Belfast International Airport or Belfast City Airport, we 
recommend that you take an airport bus to central Belfast, and then 
a bus to Derry.  The University is a short walk away, but you may 
prefer to take a taxi.  
---------------------------------------------------------------------
EXTENDED ABSTRACT COVER - IMVIP '97 
Return to: J. CAMPBELL, FACULTY OF INFORMATICS, UNIVERSITY OF ULSTER,
L'DERRY BT48 7JL, NORTHERN IRELAND. EMAIL: imvip97@ulst.ac.uk, 
TEL +44 1504 375367, FAX +44 1504 370040
1. Paper Title:_______________________________________________________
______________________________________________________________________
2. Name(s) of Author(s)
Lastname:__________________________Firstname:_________________________
Org./Dept.:___________________________E-mail:_________________________
Lastname:__________________________Firstname:_________________________
Org./Dept:____________________________E-mail:_________________________
                                Use additional sheets as neccessary.
3. Corresponding author.
Lastname:__________________________Firstname:_________________________
Address, Organisation:________________________________________________
Department:___________________________________________________________
Street Address:_______________________________________________________
City:__________________________ State:________________________________
Postal/Zip-code:_______________ Country:______________________________
E-mail:_______________________________________________________________
4. Concise description of problem addressed and its importance:
______________________________________________________________________
______________________________________________________________________
______________________________________________________________________
______________________________________________________________________
______________________________________________________________________
5. Concise statement of the originality of the contribution, plus
mention of comparison with existing work:
______________________________________________________________________
______________________________________________________________________
______________________________________________________________________
______________________________________________________________________
======================================================================
REGISTRATION INFORMATION - IMVIP '97
Registration fee schedule:  
Early registration, payable before Friday 29 August 1997:   GBP 120
Late registration:                                          GBP 150
Reduction for OESI/SPIE members: GBP 20 on either early or late
registration.
Payment by cheque to "IMVIP/AI-97, University of Ulster".  Note that 
no distinction is made in regard to the tandem conference AI'97 - the 
registration fee entitles participation at both.  For registration,
please 
use the form below.  The registration fee includes conference materials, 
coffees, and a copy of the proceedings.
Accommodation: nearby choices for hotel, guesthouse and bed and
breakfast, 
are posted on the conference's Web area, together with travel
information.
University student accommodation (price Pds 14.10 per night, breakfast
in
addition) is available also.
REGISTRATION FORM - IMVIP '97
Return to: Ms C McNUTT, FACULTY OF INFORMATICS, UNIVERSITY OF ULSTER,
L'DERRY BT48 7JL, NORTHERN IRELAND. EMAIL: imvip97@ulst.ac.uk, 
TEL +44 1504 375408, FAX +44 1504 370040
Lastname:__________________________Firstname:_________________________
Address, Organisation:________________________________________________
Department:___________________________________________________________
Street Address:_______________________________________________________
City:__________________________ State:________________________________
Postal/Zip-code:_______________ Country:______________________________
E-mail:_______________________________________________________________
PAYMENT: 
Registration fee enclosed:                                    ________
I request a room in University accommodation:                      
  Nights required:                                            ________
  Payment will be made directly on receiving key.
  See conference Web area for contact details for 
  local hotels.
I will attend the Conference banquet (GBP 25) on Friday 
  12 September 1997:                                          ________
I will attend the tour to the Giant's Causeway and to 
  Bushmills Distillery in the afternoon of Sunday
  14 September 1997 (GBP 6):                                  ________
I enclose a Sterling cheque (or Eurocheque) to the value of:
  If paying by IEP, please apply current rate of conversion 
  plus GBP 5.                                                 ________
I will attend the reception (no charge) on Thursday 11          
  September 1997:                                               yes/no
Total payment enclosed:                                       ________
======================================================================
-- 
Jonathan G. Campbell, ISC/ISE, University of Ulster, Magee College, 
Derry, BT48 7JL, Northern Ireland. tel +44 1504 375367, fax 370040.
JG.Campbell@ulst.ac.uk            http://www.iscm.ulst.ac.uk/~jon/

I have recently been studying a geostatistical method called "kriging",
named after a Mr. Krige.  None of my references give the pronunciation of
"kriging".  Does anyone out there know how pronounce this word?  Is the
"g" hard or soft?  Does the "i" sound as in "pig" or is it a long "e"
sound?  Any informed opinions welcomed.  Thanks.
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
Charles P. Reeve, golfing statistician
Disclaimer:  I speak only for myself and the guy who is beating that drum.
 ~~~~~~~~~~  "Gort, Klaatu Barada Nikto!" ... Helen ~~~~~~~~~~

If  you have two pieces of equipment that fail x1 and x2 times per year
(independently of each other), and for a given failure the failed item is
unavailable for y1 and y2 days respectively, what is the distribution for
the number of times that both items are unavailable together ?  In
particular, if you define the intersection event as any period for which
both items are unavailable, what is the expected number of intersection
events per year and the average duration of the intersection event ?
I have tried the following to estimate the average number of times that
both are unavailable :
Item	failure rate	average down 	Expected total	Fraction of year
	times per year	time per failure	time down	unavailable
			days		days/year		
A	x1		y1		x1.y1		x1.y1/365 = P(A)
B	x2		y2		x2.y2		x2.y2/365 = P(B)
A & B	c		d		e		P(AB) = P(A).P(B)
e = (x1.y1/365) . (x2.y2/365) . 365
d = min(y1,y2) . [y1+y2-min(y1,y2))] / (y1+y2)
c = d / e
I feel that the formulas for e and c are okay, and the formula for d seems
to give the rigth answers 
but I'm not convinced that there's any sound basis for the formula.  I
would also like to understand
what sort of distribution arises for the number of overlapping or combined
failures and check that the
mean number of combined failures gives the same answer as I have above.
I would welcome comments or suggestions.
Thanks
Jeff

In article <01bc72f8$6d53de00$67962fce@default> "Barb Warf"  writes:
>Does anyone know of a good source which explains the background for the
>simples method for generating the maximum values in linear programming?
>warfie
If by "background" you mean the history of the simplex method for solving 
linear programs, there is no better reference than the orignal:
   Dantzig, George B., "Linear Programming and Extensions,"
   Princeton University Press, 1963.
Chapters one and two, "The Linear Programming Concept" and "Origins and 
Influences," have no equal for historical perspective.  Although out of date 
and out of print, it should be available at any good technical library.  I 
wouldn't part with my copy (lugged around since 1967 -- price $11.50).
If by "background," however, you mean the mathematical basis for the simplex 
(and more recent) methods for solving linear programs, I recommend that you 
start with the classic text
   Hillier, Frederick S., and Lieberman, Gerald J., "Introduction to
   Operations Research," McGraw-Hill, Inc., Sixth Edition, 1995.
and use the "selected references" sprinkled through the first eight chapters.
(Yes, I've lugged their first edition around since 1968, but I don't have a 
record of the price.)  This should also be available in any good technical 
library, and at any major college book store (check under courses for 
Industrial Engineering, Operations Research, and Industrial Management).
BTW, this topic probably has a wider group of experienced followers in the 
newsgroup "sci.op-research"  Good luck!
----------------------------------------------------------------------
Charley Harp, N8MQL         FBP-425                     charp@ford.com
Operations Research Dept.   555 Republic Dr.         V: (313) 845-5873
Ford Motor Company          Allen Park, MI  48101    F: (313) 621-8381
----------------------------------------------------------------------

Hi everybody!
3 questions... I am doing some econometrics on economic data, and I
encountered some (small) problems. So if someone can give me hints or
ideas...
1) I collected my data from a relatively reliable source, but it seems to
me that the  nature of it slightly changed over time. How can I check that
my data are homogeneous over time, or, in other wors, how to identify a
breakpoint?
2) My data are globally monthly. But for one year I could only get it
quarterly. So I replaced the missing data by the forecast I got by ARIMA
modelization. Is there a way to use the information on cumumative data
(quarterly) to refine it?
3) Is it recommended to use Seasonal Exponential Smoothing on raw data, or
on stationarized series?
Thanks!
Mathias Brandewinder

BIOSTATISTICAN
Kansas Foundation for Medical Care has an immediate opening for a
biostatistican to support clincial quality improvement efforts. 
Applicants must a possess a minimum of master's degree in statistics,
epidemiology, or public health along with a strong understanding of the
health care setting.  Field experience in clinical studies, including
study design and methodology is required.  KFMC offers excellent
benefits and competitive compensation.
Please send resumes e-mail, fax, 913-273-5130 or KFMC, 2947 SW Wanamaker
Drive, Topeka, KS 66614.
-- 
*****************************
Evelyn K. Headley
Manager, Human Resources
eheadley@kfmc.org
*****************************

Could somebody please explain how one could fit a conditional
logistic regression model using S-plus? Do they fit into the
usual GLM framework?
Thanks,
Dave

Dnorlund said:
>I have to agree with Sean Ellis and disagree with Mod Tollens (although
Sean calculated probability not odds).  I read the problem as what is the
probability of turning over three cards from a 52 card deck and getting at
least one of the specified 12 cards (4 jacks, 4 sevens, 4 twos).  Sean's
approach of simplifying the problem by calculating the complement of the
desired probability is the approach I would have used as well.<
I bow my head in humility and admit my mistake.
I do think I had a mental attack of sorts -- probably an aberrant form of
dislexia.  It all seems so simple now.  I'm not so sure it would be simple
to attack the problem head on, though.  
This raises another question.  If a direct method is difficult, is the
complement always easy.  I don't think so.  If I am correct, what are we
to do with these kind of problems in which direct and indirect are both
difficult?  Computers???
                              Mod Tollens,  
                                                  --bowing again in
humility -- Please give me a more difficult problem to regain my
composure, if you will.  Thank you.

Just take a 0-1 uniform random number generator and multiply the
result by 2^32-1.
Eric
eweiss@winchendon.com

ModTollens wrote:
> 
> Such an interesting thought.  I think this allows me to ask the question,
> "Is life fair?"
> 
> I am very interested in any response.
> 
>      Mod Tollens  ......................  in the meantime, I am going to
> the Poker discussion to apologize for my error.
Sure!
Life, in the long run, is fair!!
The central limit theorem allows us to assume a normal distribution for
large number of observations of any random variable. So this means that,
in the long run, the bad stuff that happens should even out with the
good stuff that happens. Furthermore, since bad experiences and good
experiences should cancel each other out in the long run then the mean
of this Gaussian should be zero.
Does this make sense? I'm not sure...I'm just ranting at this point....
Bye,
-- 
       "In the Beginning there was nothing, which exploded."
         http://www4.ncsu.edu/eos/users/p/peguaris/WWW/
         *********************************************
         *            Pavel E. Guarisma N.           *
         *           peguaris@eos.ncsu.edu           *
         *                                           *
         *    Operations Research Graduate Program   *
         *      North Carolina State University      *
         *********************************************

I have this problem:
let X be the random variable whose probability distribution
function is the so called Cantor ternary function (constructed
using the Cantor middle-third set in [0,1]). It is a well known
example of a random variable that is continuous but not absolutely
continuous.
Find expectation and variance of X.

Hi:
            Could any one tell me a real example of application of the
law of the greater number's?
            What's the pratical use in real life of this law?
   Please send answer's by e-mail to:
                                                            etjffr@ua.pt
or foward to this question ( preference to e-mail)!
regard's and thank's
                                        José Rosado

Hi there,
	I am looking for a little assistance with remebering something. 
Some time ago I thought I saw a method for expanidng either an expression 
 or a set of data into a series (infinte I think) of sines ie
 f(x) = C1(x)Sin(V1(x)) + C2(x)Sin(V2(x)) ... etc
If anyone can tell me what this method is called or can tell me of any 
other way to find periodicity (if it exists) in a set of data points I 
would be most gratefull.
cheers
Nigel Senior
Melbourne Australia

Hi,
This is probably not the place to ask this question, but I am
desperate.  (I want to prevent a riot at the Senior Center).
A group of Scrabble players had their year-end fete.  They determined
the winners by taking the total of each person's game scores for the
year and dividing this by the number of games the person played.  Some
people only showed up for 13 games and so their total was divided by 13,
while others showed up for all 34 games and their total was divided by
34.  The people who showed up for all the games claim that this method
of determining the average scores for the year is more advantagious for
people who play the fewest games.  Is there any truth to this?
Thanks!
Carolyn
-- 
http://www.tiac.net/users/millie
"It is the ideal library, I think. Books are the liberated
spirits of men, and should be bestowed in a heaven of light and grace
and
harmonious color and sumptuous comfort, like this..."
--Mark Twain on the Millicent Library

Bob Wheeler  wrote:
> Carolyn Longworth  wrote:
>> This is probably not the place to ask this question, but I am
>> desperate.  (I want to prevent a riot at the Senior Center).
>> A group of Scrabble players had their year-end fete.  They determined
>> the winners by taking the total of each person's game scores for the
>> year and dividing this by the number of games the person played.  Some
>> people only showed up for 13 games and so their total was divided by 13,
>> while others showed up for all 34 games and their total was divided by
>> 34.  The people who showed up for all the games claim that this method
>> of determining the average scores for the year is more advantagious for
>> people who play the fewest games.  Is there any truth to this?
> Yes, but not in the way they are thinking.
>
> The average game scores of the occasional players
> will fluctuate more, and more frequently throw up
> a high average by chance.
>
> Suggest you divide each person's average score by 
> the range (largest minus smallest) of their scores.
What you have here is very similar to major league batting
averages.  In 1963 John Paciorek had a season batting
average of 1.000 while playing for Houston.  But the NL
batting title went to Tommy Davis of Los Angeles who had
a .326 batting average.  The difference was that Davis
played the whole season and had over 550 at bats while 
Paciorek played ony one game and had exactly three official
at bats.  I believe the Major League Baseball rule is 400
at bats.
My point is that the problem is not with the method, but
with the qualifications to be considered in the final
rankings.
Dick

I've compiled an archive containing Galois Field (GF) Tables for 
all prime and prime power values upto q=61. It can be downloaded
from my site http://homepages.muenchen.org/bm259225/
GALOIS FIELD (GF) TABLES - Author U.Mutlu
------------------------
Tables (Add/Sub/Mul/Div) for the following prime and prime power values.
3 Formats for each table available: VAL, POLY, IDX
Additionally: Example calculations, GFSquares, PowerCycles, Primitive Element (Pel), ...
    2   
    3   
    4   :   2^2
    5   
    7   
    8   :   2^3
    9   :   3^2
   11   
   13   
   16   :   2^4
   17   
   19   
   23   
   25   :   5^2
   27   :   3^3
   29   
   31   
   32   :   2^5
   37   
   41   
   43   
   47   
   49   :   7^2
   53   
   59   
   61   
Example: GF(p=3,m=2) -> q=9  (from file GF_03_02.TXT)
-------
GF(3^2=9) : f(x) = 2+x+x^2
pel = x
pelIdx = 1
iMaxStrLen = 4
Non-Zero elements:
    0: 1
    1: x
    2: 1+2x
    3: 2+2x
    4: 2
    5: 2x
    6: 2+x
    7: 1+x
2 * 1 = 2                   <--- Example Calcs
2 * 2 = 1
2 * 0 = 2+2x
2 * 1 = 2
2 * 2 = 1
2 * 0 = 2+2x
2 * 1 = 2
2 * 2 = 1
2 * 0 = 2+2x
2 * 1 = 2
2 * 2 = 1
2x + 1+x = 1
2x - 1+x = 2+x
2x * 1+x = 2
2x / 1+x = 2+x
GF(p=3,m=2,->q=9)  ADDITION Table (incl. zero-element). Fmt: VAL :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    1    2    3    4    5    6    7    8 
  1 =    1 =   1  =   0 .    1    2    0    4    5    3    7    8    6 
  2 =    2 =   3  =   4 .    2    0    1    5    3    4    8    6    7 
  3 =    x =   7  =   1 .    3    4    5    6    7    8    0    1    2 
  4 =  1+x =   8  =   7 .    4    5    3    7    8    6    1    2    0 
  5 =  2+x =   2  =   6 .    5    3    4    8    6    7    2    0    1 
  6 =   2x =   6  =   5 .    6    7    8    0    1    2    3    4    5 
  7 = 1+2x =   5  =   2 .    7    8    6    1    2    0    4    5    3 
  8 = 2+2x =   4  =   3 .    8    6    7    2    0    1    5    3    4 
GF(p=3,m=2,->q=9)  MULTIPLICATION Table (incl. zero-element). Fmt: VAL :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    0    0    0    0    0    0    0    0 
  1 =    1 =   1  =   0 .    0    1    2    3    4    5    6    7    8 
  2 =    2 =   3  =   4 .    0    2    1    6    8    7    3    5    4 
  3 =    x =   7  =   1 .    0    3    6    7    1    4    5    8    2 
  4 =  1+x =   8  =   7 .    0    4    8    1    5    6    2    3    7 
  5 =  2+x =   2  =   6 .    0    5    7    4    6    2    8    1    3 
  6 =   2x =   6  =   5 .    0    6    3    5    2    8    7    4    1 
  7 = 1+2x =   5  =   2 .    0    7    5    8    3    1    4    2    6 
  8 = 2+2x =   4  =   3 .    0    8    4    2    7    3    1    6    5 
GF(p=3,m=2,->q=9)  SUBTRACTION Table (incl. zero-element). Fmt: VAL :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    2    1    6    8    7    3    5    4 
  1 =    1 =   1  =   0 .    1    0    2    7    6    8    4    3    5 
  2 =    2 =   3  =   4 .    2    1    0    8    7    6    5    4    3 
  3 =    x =   7  =   1 .    3    5    4    0    2    1    6    8    7 
  4 =  1+x =   8  =   7 .    4    3    5    1    0    2    7    6    8 
  5 =  2+x =   2  =   6 .    5    4    3    2    1    0    8    7    6 
  6 =   2x =   6  =   5 .    6    8    7    3    5    4    0    2    1 
  7 = 1+2x =   5  =   2 .    7    6    8    4    3    5    1    0    2 
  8 = 2+2x =   4  =   3 .    8    7    6    5    4    3    2    1    0 
GF(p=3,m=2,->q=9)  DIVISION Table (incl. zero-element). Fmt: VAL :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    0    0    0    0    0    0    0    0 
  1 =    1 =   1  =   0 .    0    1    2    4    3    7    8    5    6 
  2 =    2 =   3  =   4 .    0    2    1    8    6    5    4    7    3 
  3 =    x =   7  =   1 .    0    3    6    1    7    8    2    4    5 
  4 =  1+x =   8  =   7 .    0    4    8    5    1    3    7    6    2 
  5 =  2+x =   2  =   6 .    0    5    7    6    4    1    3    2    8 
  6 =   2x =   6  =   5 .    0    6    3    2    5    4    1    8    7 
  7 = 1+2x =   5  =   2 .    0    7    5    3    8    2    6    1    4 
  8 = 2+2x =   4  =   3 .    0    8    4    7    2    6    5    3    1 
GF(p=3,m=2,->q=9)  ADDITION Table (incl. zero-element). Fmt: POLY :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
  1 =    1 =   1  =   0 .    1    2    0  1+x  2+x    x 1+2x 2+2x   2x 
  2 =    2 =   3  =   4 .    2    0    1  2+x    x  1+x 2+2x   2x 1+2x 
  3 =    x =   7  =   1 .    x  1+x  2+x   2x 1+2x 2+2x    0    1    2 
  4 =  1+x =   8  =   7 .  1+x  2+x    x 1+2x 2+2x   2x    1    2    0 
  5 =  2+x =   2  =   6 .  2+x    x  1+x 2+2x   2x 1+2x    2    0    1 
  6 =   2x =   6  =   5 .   2x 1+2x 2+2x    0    1    2    x  1+x  2+x 
  7 = 1+2x =   5  =   2 . 1+2x 2+2x   2x    1    2    0  1+x  2+x    x 
  8 = 2+2x =   4  =   3 . 2+2x   2x 1+2x    2    0    1  2+x    x  1+x 
GF(p=3,m=2,->q=9)  MULTIPLICATION Table (incl. zero-element). Fmt: POLY :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    0    0    0    0    0    0    0    0 
  1 =    1 =   1  =   0 .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
  2 =    2 =   3  =   4 .    0    2    1   2x 2+2x 1+2x    x  2+x  1+x 
  3 =    x =   7  =   1 .    0    x   2x 1+2x    1  1+x  2+x 2+2x    2 
  4 =  1+x =   8  =   7 .    0  1+x 2+2x    1  2+x   2x    2    x 1+2x 
  5 =  2+x =   2  =   6 .    0  2+x 1+2x  1+x   2x    2 2+2x    1    x 
  6 =   2x =   6  =   5 .    0   2x    x  2+x    2 2+2x 1+2x  1+x    1 
  7 = 1+2x =   5  =   2 .    0 1+2x  2+x 2+2x    x    1  1+x    2   2x 
  8 = 2+2x =   4  =   3 .    0 2+2x  1+x    2 1+2x    x    1   2x  2+x 
GF(p=3,m=2,->q=9)  SUBTRACTION Table (incl. zero-element). Fmt: POLY :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    2    1   2x 2+2x 1+2x    x  2+x  1+x 
  1 =    1 =   1  =   0 .    1    0    2 1+2x   2x 2+2x  1+x    x  2+x 
  2 =    2 =   3  =   4 .    2    1    0 2+2x 1+2x   2x  2+x  1+x    x 
  3 =    x =   7  =   1 .    x  2+x  1+x    0    2    1   2x 2+2x 1+2x 
  4 =  1+x =   8  =   7 .  1+x    x  2+x    1    0    2 1+2x   2x 2+2x 
  5 =  2+x =   2  =   6 .  2+x  1+x    x    2    1    0 2+2x 1+2x   2x 
  6 =   2x =   6  =   5 .   2x 2+2x 1+2x    x  2+x  1+x    0    2    1 
  7 = 1+2x =   5  =   2 . 1+2x   2x 2+2x  1+x    x  2+x    1    0    2 
  8 = 2+2x =   4  =   3 . 2+2x 1+2x   2x  2+x  1+x    x    2    1    0 
GF(p=3,m=2,->q=9)  DIVISION Table (incl. zero-element). Fmt: POLY :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .    0    0    0    0    0    0    0    0    0 
  1 =    1 =   1  =   0 .    0    1    2  1+x    x 1+2x 2+2x  2+x   2x 
  2 =    2 =   3  =   4 .    0    2    1 2+2x   2x  2+x  1+x 1+2x    x 
  3 =    x =   7  =   1 .    0    x   2x    1 1+2x 2+2x    2  1+x  2+x 
  4 =  1+x =   8  =   7 .    0  1+x 2+2x  2+x    1    x 1+2x   2x    2 
  5 =  2+x =   2  =   6 .    0  2+x 1+2x   2x  1+x    1    x    2 2+2x 
  6 =   2x =   6  =   5 .    0   2x    x    2  2+x  1+x    1 2+2x 1+2x 
  7 = 1+2x =   5  =   2 .    0 1+2x  2+x    x 2+2x    2   2x    1  1+x 
  8 = 2+2x =   4  =   3 .    0 2+2x  1+x 1+2x    2   2x  2+x    x    1 
GF(p=3,m=2,->q=9)  ADDITION Table (incl. zero-element). Fmt: IDX :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .   -1    0    4    1    7    6    5    2    3 
  1 =    1 =   1  =   0 .    0    4   -1    7    6    1    2    3    5 
  2 =    2 =   3  =   4 .    4   -1    0    6    1    7    3    5    2 
  3 =    x =   7  =   1 .    1    7    6    5    2    3   -1    0    4 
  4 =  1+x =   8  =   7 .    7    6    1    2    3    5    0    4   -1 
  5 =  2+x =   2  =   6 .    6    1    7    3    5    2    4   -1    0 
  6 =   2x =   6  =   5 .    5    2    3   -1    0    4    1    7    6 
  7 = 1+2x =   5  =   2 .    2    3    5    0    4   -1    7    6    1 
  8 = 2+2x =   4  =   3 .    3    5    2    4   -1    0    6    1    7 
GF(p=3,m=2,->q=9)  MULTIPLICATION Table (incl. zero-element). Fmt: IDX :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .   -1   -1   -1   -1   -1   -1   -1   -1   -1 
  1 =    1 =   1  =   0 .   -1    0    4    1    7    6    5    2    3 
  2 =    2 =   3  =   4 .   -1    4    0    5    3    2    1    6    7 
  3 =    x =   7  =   1 .   -1    1    5    2    0    7    6    3    4 
  4 =  1+x =   8  =   7 .   -1    7    3    0    6    5    4    1    2 
  5 =  2+x =   2  =   6 .   -1    6    2    7    5    4    3    0    1 
  6 =   2x =   6  =   5 .   -1    5    1    6    4    3    2    7    0 
  7 = 1+2x =   5  =   2 .   -1    2    6    3    1    0    7    4    5 
  8 = 2+2x =   4  =   3 .   -1    3    7    4    2    1    0    5    6 
GF(p=3,m=2,->q=9)  SUBTRACTION Table (incl. zero-element). Fmt: IDX :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .   -1    4    0    5    3    2    1    6    7 
  1 =    1 =   1  =   0 .    0   -1    4    2    5    3    7    1    6 
  2 =    2 =   3  =   4 .    4    0   -1    3    2    5    6    7    1 
  3 =    x =   7  =   1 .    1    6    7   -1    4    0    5    3    2 
  4 =  1+x =   8  =   7 .    7    1    6    0   -1    4    2    5    3 
  5 =  2+x =   2  =   6 .    6    7    1    4    0   -1    3    2    5 
  6 =   2x =   6  =   5 .    5    3    2    1    6    7   -1    4    0 
  7 = 1+2x =   5  =   2 .    2    5    3    7    1    6    0   -1    4 
  8 = 2+2x =   4  =   3 .    3    2    5    6    7    1    4    0   -1 
GF(p=3,m=2,->q=9)  DIVISION Table (incl. zero-element). Fmt: IDX :
lfd                     .    0    1    2    3    4    5    6    7    8 
    = poly              .    0    1    2    x  1+x  2+x   2x 1+2x 2+2x 
           = val        .    0    1    3    7    8    2    6    5    4 
                  = idx .   -1    0    4    1    7    6    5    2    3 
......................................................................
  0 =    0 =   0  =  -1 .   -1   -1   -1   -1   -1   -1   -1   -1   -1 
  1 =    1 =   1  =   0 .   -1    0    4    7    1    2    3    6    5 
  2 =    2 =   3  =   4 .   -1    4    0    3    5    6    7    2    1 
  3 =    x =   7  =   1 .   -1    1    5    0    2    3    4    7    6 
  4 =  1+x =   8  =   7 .   -1    7    3    6    0    1    2    5    4 
  5 =  2+x =   2  =   6 .   -1    6    2    5    7    0    1    4    3 
  6 =   2x =   6  =   5 .   -1    5    1    4    6    7    0    3    2 
  7 = 1+2x =   5  =   2 .   -1    2    6    1    3    4    5    0    7 
  8 = 2+2x =   4  =   3 .   -1    3    7    2    4    5    6    1    0 
Power cycle table of pel:
x ^0=1
x ^1=x
x ^2=1+2x
x ^3=2+2x
x ^4=2
x ^5=2x
x ^6=2+x
x ^7=1+x
x ^8=1
PowerCycle2 of pel:
x ^0=1
x ^1=3
x ^2=7
x ^3=8
x ^4=2
x ^5=6
x ^6=5
x ^7=4
x ^8=1
GFSquares (incl. 0):
   0  ^2 =    0 
   1  ^2 =    1 
   x  ^2 = 1+2x 
1+2x  ^2 =    2 
2+2x  ^2 =  2+x 
   2  ^2 =    1 
  2x  ^2 = 1+2x 
 2+x  ^2 =    2 
 1+x  ^2 =  2+x 
### end ###
--
NEW email adress and homepage location with some interessting files and links:
U.Mutlu (bm259225@muenchen.org)  http://homepages.muenchen.org/bm259225/
in Munich & Hannover / Germany

Can someone send me the statistical formula for establishing the cost
effectiveness of using a psychometric test?

In <1997Jun19.092626.9124@srs.gov> charles.reeve@srs.gov (Charles P.
Reeve) writes: 
> None of my references give the pronunciation of
>"kriging".  Does anyone out there know how pronounce this word?  Is
the
>"g" hard or soft?  Does the "i" sound as in "pig" or is it a long "e"
>
The 'g' is hard and the 'i' is the long 'e.'
'kreeging'
Sue Hawkins

Can someone send me the statistical formula for establishing the cost
effectiveness of using a psychometric test?

Yes, but not in the way they are thinking.
The average game scores of the occasional players
will fluctuate more, and more frequently throw up
a high  average by chance.
Suggest you divide each person's average score by 
the range (largest minus smallest) of their scores.
-- 
Bob Wheeler, ECHIP, Inc. 
    Reply to bwheeler@echip.com)
Carolyn Longworth  wrote in article
<33ABD26A.7332@ultranet.com>...
> Hi,
> This is probably not the place to ask this question, but I am
> desperate.  (I want to prevent a riot at the Senior Center).
> A group of Scrabble players had their year-end fete.  They determined
> the winners by taking the total of each person's game scores for the
> year and dividing this by the number of games the person played.  Some
> people only showed up for 13 games and so their total was divided by 13,
> while others showed up for all 34 games and their total was divided by
> 34.  The people who showed up for all the games claim that this method
> of determining the average scores for the year is more advantagious for
> people who play the fewest games.  Is there any truth to this?
> Thanks!
> Carolyn
> -- 
> http://www.tiac.net/users/millie
> 
> "It is the ideal library, I think. Books are the liberated
> spirits of men, and should be bestowed in a heaven of light and grace
> and
> harmonious color and sumptuous comfort, like this..."
> --Mark Twain on the Millicent Library
> 
> 

please help a damsel (no, make that damson) in distress!
am stuck with a casio fx-85b calc without r button, and casio fx-100c with
r button but minus instructions.  Evolution course requires some heavy (at
least to me) calculations and essay requires r for 2 sets of data in SD.
Please - does anyone know how to work the damn things?
yours in desperation
D Wilson

The closing date for the previously posted position is of course, 
25th July 1997, not 1998 as previously stated! Monday Morning......
-- 
 Glenn Stone
 Statistician, CSIRO
 Locked Bag 17, North Ryde, NSW 2113
 Phone:+61 2 9325 3216, Fax:+61 2 9325 3200
 Glenn.Stone@cmis.csiro.au
 http://www.dms.CSIRO.AU/~gstone

Can anybody please explain the differences between hazard functions and
PROBIT/LOGIT? 
-- 
Jefferson Glapski
Visit the webpage of the above at: http://home.ican.net/~jng/

Dick Adams wrote:
> 
> Bob Wheeler  wrote:
> > Carolyn Longworth  wrote:
> 
> >> A group of Scrabble players had their year-end fete.  They determined ....
 ople who play the fewest games.  Is there any truth to this?
> 
> > Yes, but not in the way they are thinking.
> >
> > The average game scores of the occasional players
> > will fluctuate more, and more frequently throw up
> > a high average by chance.
> >
> > Suggest you divide each person's average score by
> > the range (largest minus smallest) of their scores.
> 
> What you have here is very similar to major league batting
> averages.  
>
> I believe the Major League Baseball rule is 400 at bats.
Perhaps a technical way out is to use Stein shrinkage. In effect
everybodies scores are pulled towards the group-average, but less
so as their number of games increases.
A technically defensible approach, but I doubt that the committee
would adopt it.
Peter

On 19 Jun 1997, Joern Apel wrote:
> who could please give me a quick hint which procedure is used to compute
> the following: I have got a series of measures varying about a certain 
> value. say: 48,49,43,51,47, 49,48,...etc. now there are a few numbers 
> completely out of range due to systematic errors, say: 1,2 and 178,211 
> (let me call them "tear out's" :-) )
They are known as "outliers".
> If I plot the histogram, I can see those concentratet in intervals at both
> ends of the range and the ones I want to extract are well distributet in
> the middle of it. How can I find (compute) the intervals to delete and 
> the intervals to leave in order to proceed with statistics on them?
> any (tiny) hint is very much appreciated (just the name of any test is
> fine),
I'm not sure if Dixon's Q statistic is appropriate. But you can check it out.
Also check the following books:
Barnett and Lewis "Outliers in Statistical Data", Wiley, 1984.
Hawkins, "Identification of Outliers", Chapman and Hall, 1980.
Huber, "Robust Statistical Procedures", SIAM, 1977.
Hope this helps.
Gregory E. Heath     heath@ll.mit.edu      The views expressed here are
M.I.T. Lincoln Lab   (617) 981-2815        not necessarily shared by
Lexington, MA        (617) 981-0908(FAX)   M.I.T./LL or its sponsors
02173-9185, USA

Gary Robinson wrote:
> 
> Hello,
> 
> Does anyone know of a particularly good algorithm for generating a 32-bit
> or larger random number?
Abramovitz and Stegun, chapter 26: probability functions.  Several
multiplicative congruential generators up to 40 bits whose properties
have been tested.
JEH

I am comparing two versions of an ASTM specification (D2992).  One uses a
"linear functional relationship" (covariance) and one uses the linear
least squares method to calculate information about a curve fit line for
test data.
It appears that the method for calculating the slope of this line is
different for these two.  Is this correct?
I apologize if this is a basic question.  I can't seem to find an answer
in any of my engineering or science textbooks.	Any assistance is greatly
appreciated!
-------------------==== Posted via Deja News ====-----------------------
      http://www.dejanews.com/     Search, Read, Post to Usenet

Postdoctoral Fellowship
CSIRO Mathematical & Information Sciences
North Ryde  NSW  Australia
Postdoctoral Fellowship - Term 3 years
$AUS 41,000 - 47,000 + superannuation
We wish to appoint a Post-Doctoral fellow to join a research team
working on large and complex datasets. Your PhD in statistics,
computer science, or related discipline or equivalent must have been
awarded with the last three years.
The team consists of Statisticians and Computer Scientists with
interest in techniques for handling and cleaning large datasets,
methods for modeling large datasets, wavelet methods for feature
extraction, statistical visualisation and modeling multiple time
series.  The team is working on datasets coming from areas as diverse
as motor vehicle insurance, finance, marketing and astronomy.
The project would suit an applicant with experience analysing
real-world datasets. You will need excellent computing skills in C or
C++, or a statistical package such as S-Plus or SAS. Ability to work
in a team and demonstrated ability to meet deadlines. 
The position is for a term of three (3) years.  Further information
about the position may be obtained from 
 Dr Glenn Stone, tel +61 2 9325 3216 email: glenn.stone@cmis.csiro.au  
The job description and selection criteria may be obtained from 
 Lucinda Wells, tel +61 2 9325 3277  email: lucinda.wells@cmis.csiro.au 
Applications for the position should address the selection criteria,
be marked "Confidential" quoting reference number MS 97/1, and be 
sent to: The Human Resources Manager, CSIRO, Division of Mathematical
and Information Sciences, Locked Bag 17, North Ryde  NSW  2113 
by 25th July, 1998. 
-- 
 Glenn Stone
 Statistician, CSIRO
 Locked Bag 17, North Ryde, NSW 2113
 Phone:+61 2 9325 3216, Fax:+61 2 9325 3200
 Glenn.Stone@cmis.csiro.au
 http://www.dms.CSIRO.AU/~gstone

Hello,
I was wondering if someone could figure out a relativley simple
equation. 
I need to know the probabilties for a deck of cards - but not an
ordinary deck, a five suited deck of cards.
What I need to know is the probability of 5 suits and 65 cards.
There would be additions to the example list below - mainly 5 of-a-kind
(where would this fall).
---------------------------
This is an example of a 52 card deck and its outcomes:
The standard poker hands are ranked based on the probability of their
being
dealt pat in 5 cards from a full 52-card deck. The following table lists
the hands in order of increasing frequency, and shows how many ways each
hand can be dealt in 3, 5, and 7 cards.
Hand                  3 cards           5 cards           7 cards
====                  =======           =======           =======
Straight Flush             48                40            41,584
Four of a Kind              0               624           224,848
Full House                  0             3,744         3,473,184
Flush                   1,096             5,108         4,047,644
Straight                  720            10,200         6,180,020
Three of a Kind            52            54,912         6,461,620
Two Pair                    0           123,552        31,433,400
One Pair                3,744         1,098,240        58,627,800
High Card              16,440         1,302,540        23,294,460
=================================================================
TOTALS                 22,100         2,598,960       133,784,560
Notes:
1. The standard rankings are incorrect for 3-card hands, since it is
easier
to get a flush than a straight, and easier to get a straight than three
of
a kind.
2. For 7-card hands, the numbers reflect the best possible 5-card hand
out
of the 7 cards. For instance, a hand that contains both a straight and
three of a kind is counted as a straight.
3. For 7-card hands, only five cards need be in sequence to make a
straight, or of the same suit to make a flush. In a 3-card hand a
sequence
of three is considered a straight, and three of the same suit a flush.
These rules reflect standard poker practice.
4. In a 7-card hand, it is easier for one's *best* 5 cards to have one
or
two pair than no pair. (Good bar bet opportunity!) However, if we
changed
the ranking to value no pairs above two pairs, all of the one pair hands
and most of the two pair hands would be able to qualify for "no pair" by
choosing a different set of five cards.
5. Within each type of hand (e.g., among all flushes) the hands are
ranked
according to an arbitrary scheme, unrelated to probability. 
Well, thanks if anyone can help me out!
Sean Rice
srice@interstyle.co

I have two (for the sake of argument) *closed* loops defined by a set of
points for each loop. What I want to do is to form the union of the two
regions enclosed by the two loops, such that if the loops do not cross at
all, then the union will just be the two loops. If however they DO cross,
then I want a single loop enclosing the union.
Is the best way to go about it to find the crossing points of the two loops
and to cut out the bits of the loop in between? If so, what is the best way
of doing this? If not, what method should I use?
Thanks,
Mike

Domenico Colucci wrote:
> 
> I have this problem:
> let X be the random variable whose probability distribution
> function is the so called Cantor ternary function (constructed
> using the Cantor middle-third set in [0,1]). It is a well known
> example of a random variable that is continuous but not absolutely
> continuous.
> Find expectation and variance of X.
X can be expressed as the sum as i goes from 1 to infinity of
Y_i/3^i where the Y_i's are i i d and take the value
0 with probability 1/2 and 2 with probability 1/2. E(Y_i) = 1
and Var(Y_i) = 1. It follows that E(X)= sum(i=1 to inf) 1/3^i = 1/2
and Var(Y) = sum(i=1 to inf) 1/9^i=1/8.
Hope this helps.
Ellen Hertz
E(X) =

In <5og90u$kgs$3@eplet.mira.net.au> nige@werple03.mira.net.au (Nigel
Senior) writes: 
>
>Hi there,
>	I am looking for a little assistance with remebering something. 
>Some time ago I thought I saw a method for expanidng either an
expression or a set of data into a series (infinte I think) of sines ie
>
> f(x) = C1(x)Sin(V1(x)) + C2(x)Sin(V2(x)) ... etc
>
>If anyone can tell me what this method is called or can tell me of any
other way to find periodicity (if it exists) in a set of data points I 
would be most gratefull.
>
>
>cheers
>
>Nigel Senior
>Melbourne Australia
>
Hi,
The usual way to do this is to compute the power spectrum of the data.
The peaks of the spectrum locate the frequencies that dominate your
data. This can be done with most statistical packages. I recommend
SPSS/BMDP CLASSIC.
Once you have the major frequencies you can regress the original data
against sines and cosines at the given frequencies as independent
variables within multivariate linear regression. Again, most
statistical packages have this capability.
Cheers, Tony Badalamenti

Carolyn Longworth  wrote:
Date: Sat, 21 Jun 1997 09:08:58 -0400
Message-ID: <33ABD26A.7332@ultranet.com>
>Hi,
>This is probably not the place to ask this question, but I am
>desperate.  (I want to prevent a riot at the Senior Center).
>A group of Scrabble players had their year-end fete.  They determined
>the winners by taking the total of each person's game scores for the
>year and dividing this by the number of games the person played.  Some
>people only showed up for 13 games and so their total was divided by 13,
>while others showed up for all 34 games and their total was divided by
>34.  The people who showed up for all the games claim that this method
>of determining the average scores for the year is more advantagious for
>people who play the fewest games.  Is there any truth to this?
>Thanks!
>Carolyn
>-- 
>http://www.tiac.net/users/millie
Well, Carolyn as others have noted, yes, there is some truth to this.
(Imagine a poor player that was unusually lucky for several games, then
left for the rest of the "season")
A possible "solution" to propose might be that the "hard-core" players
could choose from their best 13 games - to even things up.
I imagine *no-one* will like this - and they'll start working on a
more democratic method.
Mike Sharkey
MnPSharkey@aol.com

simw120.zip - Simstat v1.20: Statistical analysis program
I have uploaded to Simtel.Net:
http://www.simtel.net/pub/simtelnet/win3/math/simw120.zip
ftp://ftp.simtel.net/pub/simtelnet/win3/math/simw120.zip   2650641 bytes
simw120.zip     Simstat v1.20: Statistical analysis program
Simstat is a statistical program providing a wide range of statistical
features such as frequency analysis cross tabulation, inter-raters
agreement statistics, t-test, ANOVA/ANCOVA, correlations, linear,
nonlinear and multiple regression, time-series, reliability and
sensitivity analysis (ROC curve), single case experimental design,
several nonparametric tests, and bootstrap resampling techniques.
The program imports and exports to most spreadsheet and database file
formats and produce presentation quality tables and graphs.  An
integrated macro language allows one to automate statistical analysis,
create interactive tutorials, demonstration programs, and even create
computer assisted interviewing programs.
Changes: Now features multilingual interface (English, French and Dutch).
         Much improved computation speed (from 200% to 800% faster).
         Conditional strings computation.  Customizable menu system.
         Interactive partial correlation module.
Special requirements: None.
simw120.zip has replaced simw112.zip.
Shareware.  Uploaded by the author.
Normand Peladeau, Provalis Research
simstat@compuserve.com
CIS: 71760,2103
http://ourworld.compuserve.com/homepages/simstat/