--- In Statisticians_group@..., "swannies1983" <split_city@...> wrote:
>
> Hi there,
>
> I ran a study in which I investigated changes in several variables leading into upper airway obstruction in sleep apnoea patients. Sleep apnoea is a condition in which the airway closes (apnoeas) repetitively throughout the night.
>
> I studied six sleep apnoea patients. Each patient had a variable number of apnoeas for analysis. One of the variables I
> looked at was gastric pressure (Pga) at end expiration. Here is a link
> showing the mean Pga for the last four breaths leading into upper airway collapse for the six patients http://users.adam.com.au/swans1983/Gastric%20Pressure.jpg The left most point is average awake data. Fot this data, I averaged about 30-40 breaths of Pga data from wakefulness in each subject. Please also see the left most data of the "Pga Pressure Data" excel file located here http://users.adam.com.au/swans1983/ The factors and levels are as follows:
>
> 1) Subject (six patients)
> 3) Apnoea# (Variable between subjects)
> 4) Breath# (Four breaths. Not all apnoeas have four breaths)
> 5) Gastric pressure (% of wake) *I have shown absolute values in the graph but for the purpose of my analysis, I have defined all breaths as a % of wake.
>
> I wish to determine whether each of the four breaths prior to apnoeas is different from wake. I have been trying to use linear mixed models in SPSS for this. Is there a syntax I can use to determine if each of the four breaths is different from 100(%)?
>
> I also thought of another method which could be used to determine if those four breaths are different from wake. Firstly, I decided to use only 30 wake breaths for each subject. So now I have:
>
> 1) Stage (0=wake, 1=apnoea)
> 2) Subject (six patients)
> 3) Period# (only 1 for wake stage, but many for apnoea stage. Variable
> numbers for each subject for apnoea stage)
> 4) Breath# (30 breaths for wake data and 4 breaths for apnoea stage. Not all apnoeas had four breaths)
> 5) Gastric pressure (absolute values)
>
> This is the syntax I tried using:
>
> MIXED Pga BY Stage Period Breath#
> /CRITERIA=CIN(95) MXITER(100) MXSTEP(5) SCORING(1)
> SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE)
> PCONVERGE(0.000001, ABSOLUTE)
> /FIXED=Stage Breath# Stage*Breath# | SSTYPE(3)
> /METHOD=REML
> /RANDOM=INTERCEPT | SUBJECT(Subject) COVTYPE(VC)
> /REPEATED=Stage*Period*Breath# | SUBJECT(Subject) COVTYPE(AR1)
> /EMMEANS=TABLES(Stage) COMPARE ADJ(SIDAK)
> /EMMEANS=TABLES(Breath#) COMPARE ADJ(SIDAK)
> /EMMEANS=TABLES(Stage*Breath#)
> /TEST = "average of wake vs breath 1 of apnoea" Stage*Breath# 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 -1 0 0 0.
>
> I thought this test would compare breath#1 of stage 1 with the average value of the 30 breaths of stage 0. I would then change the last bit of the code for the other 3 breaths. However, when I run the model, I get the error message
>
> "Custom hypothesis test 1 (average of wake vs breath 1 of apnoea) will not be performed because the L matrix is not estimable."
>
> Any ideas what I am doing wrong?

I was able to fix the the above problem via different coding but I am still interested in how to determine if breath data is different from 100% as described in my first part of my post.

Dear all

I am using AMOS for my data analysis. My data is not fit with my model and model
need modifications.Will anybody highlight willingness and help on this area so
that I can complete my analysis

lara

Hi there,

I ran a study in which I investigated changes in several variables leading into
upper airway obstruction in sleep apnoea patients. Sleep apnoea is a condition
in which the airway closes (apnoeas) repetitively throughout the night.

I studied six sleep apnoea patients. Each patient had a variable number of
apnoeas for analysis. One of the variables I
looked at was gastric pressure (Pga) at end expiration. Here is a link
showing the mean Pga for the last four breaths leading into upper airway
collapse for the six patients
http://users.adam.com.au/swans1983/Gastric%20Pressure.jpg The left most point is
average awake data. Fot this data, I averaged about 30-40 breaths of Pga data
from wakefulness in each subject. Please also see the left most data of the "Pga
Pressure Data" excel file located here http://users.adam.com.au/swans1983/ The
factors and levels are as follows:

1) Subject (six patients)
3) Apnoea# (Variable between subjects)
4) Breath# (Four breaths. Not all apnoeas have four breaths)
5) Gastric pressure (% of wake) *I have shown absolute values in the graph but
for the purpose of my analysis, I have defined all breaths as a % of wake.

I wish to determine whether each of the four breaths prior to apnoeas is
different from wake. I have been trying to use linear mixed models in SPSS for
this. Is there a syntax I can use to determine if each of the four breaths is
different from 100(%)?

I also thought of another method which could be used to determine if those four
breaths are different from wake. Firstly, I decided to use only 30 wake breaths
for each subject. So now I have:

1) Stage (0=wake, 1=apnoea)
2) Subject (six patients)
3) Period# (only 1 for wake stage, but many for apnoea stage. Variable
numbers for each subject for apnoea stage)
4) Breath# (30 breaths for wake data and 4 breaths for apnoea stage. Not all
apnoeas had four breaths)
5) Gastric pressure (absolute values)

This is the syntax I tried using:

MIXED Pga BY Stage Period Breath#
   /CRITERIA=CIN(95) MXITER(100) MXSTEP(5) SCORING(1)
SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE)
PCONVERGE(0.000001, ABSOLUTE)
   /FIXED=Stage Breath# Stage*Breath# | SSTYPE(3)
   /METHOD=REML
   /RANDOM=INTERCEPT | SUBJECT(Subject) COVTYPE(VC)
   /REPEATED=Stage*Period*Breath# | SUBJECT(Subject) COVTYPE(AR1)
   /EMMEANS=TABLES(Stage) COMPARE ADJ(SIDAK)
   /EMMEANS=TABLES(Breath#) COMPARE ADJ(SIDAK)
   /EMMEANS=TABLES(Stage*Breath#)
/TEST = "average of wake vs breath 1 of apnoea" Stage*Breath# 1/30 1/30 1/30
1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30
1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 1/30 -1 0 0 0.

I thought this test would compare breath#1 of stage 1 with the average value of
the 30 breaths of stage 0. I would then change the last bit of the code for the
other 3 breaths. However, when I run the model, I get the error message

"Custom hypothesis test 1 (average of wake vs breath 1 of apnoea) will not be
performed because the L matrix is not estimable."

Any ideas what I am doing wrong?

Thanks

Daniel

Hi everyone,

I want to call a function of R (cov.mcd which is in the package MASS of R) in a
matlab function. What I mean is the following.

Suppose I am writing a program in matlab and I want to use this particular R
function cov.mcd(dataset) within the matlab program.

If anyone can help me in this issue I will be grateful.

Thanks and Regards,

Satyaki

hi
pls anyone knows about mixed paired &unpaired analysis?

Hello,

My name is Tracy Wolfe, Vice-President of Zingaro & Company, an executive search
consulting firm specializing in the search and selection of executives for the
healthcare industry.

I am currently working on an exciting assignment with my client, an
international contract research organization with offices in India and Chicago,
IL.  We are looking for dynamic individuals interested in joining the
Statistics/SAS Programming team.  We seek candidates with a Master's degree, PhD
preferred, and a minimum of 5 years of experience in statistics, biostatistics
or a closely related field.  Additionally, we are looking for candidates with
SAS programming experience.  Individuals who are creative and thrive in an
evolving environment should consider this opportunity.

I would be grateful for your assistance in our efforts.  Please let me know of
any outstanding candidates who may be interested in this opportunity or have
them contact me directly.

Respectfully,

Tracy Wolfe
Vice President
____________________________________________
Zingaro & Company
21936 Briarcliff Drive - Briarcliff, TX 78669-2012
Phone (512) 327-7275 - Fax (512) 327-1774
www.zingaro.com
____________________________________________


Job Profile
Senior Statistician

The Company:
CliniRx Research is a part of JK Organization.  JK is a 125 year old, $2.5
billion enterprise with presence in over 40 countries and leadership in multiple
sectors.

Over the past several years, CliniRx Research has become one of the fastest
growing CROs in the country. The group's passion for excellence and its
commitment to the cause of human advancement, create an atmosphere for
achievement in the global research business.

Scope of the Job:
This new position, located at US company headquarters in Chicago, IL reports to
the Manager of Statistics.  You will serve as the lead statistician for the
analysis of clinical trials being conducted by the company.  You will interact
with the data management department, the clinical development department, and
medical writing.  Additionally, you will interact with the client to carry out
the clinical trial.  This job requires knowledge of statistics in clinical
trials, SAS programming, good communication skills, and excellent attention to
detail.

Responsibilities:
1. You will serve as the lead statistician on project teams for the execution of
clinical trials.  This includes:
­ Regular contact with the client.
­ Coordinating the planning and execution of the statistics related tasks as per
the contract.
­ You will work with the project team to establish a clear project scope and
timeline.
­ Working with the study data manager to ensure project deliverables are
achieved and in compliance with regulatory requirements.
­ You will ensure high quality, on time statistical deliverables.
­ Managing the financial aspects for the statistical portion of the project.
­ Working closely with the programmers to coordinate deliverables.

2. Preparation and review of statistical analysis plans in accordance with the
study protocol for more complicated or unconventional clinical trials.
3. Writes and reviews statistical method section for protocols independently.
4. Estimates sample sizes for clinical trials.
5. Advises clients on statistical design issues.
6. Performs QC of tables, listings, graphs, and derived datasets, particularly
those of a more complicated nature or those used for efficacy analyses.
7. Reviews CRFs and data management plans.
8. Can independently develop computer programs to generate tables, listings, and
graphs using SAS as outlined by the statistical analysis plan.   Candidates must
have an in-depth understanding of p-values.
9. Performs quality review of statistics deliverables to clients.
10. Works with programmers to ensure statistical analyses are accurate and
carried out correctly.
11. Reviews clinical study reports.
12. May assist in business development activities including study pricing and
new business acquisition.
13. Adheres to all study/project timelines.
14. Documents work so other statisticians and programmers can check or modify
work.
15. Suggests process improvements for work procedures.
16. Assists with the training and mentoring of new employees as needed.
17. Adheres to all policies, procedures, SOPs, and training plans.
18. Leads and/or contributes to internal process improvement initiatives.
19. Files study documents into Biometrics internal files.
20. Reviews draft protocols and CRFs for potential data collection and
representation, or database structure problems.  Provides feedback to the
project team.
21. Other duties as assigned.


Candidate Requirements:
We seek a statistician with a minimum of 5 years of experience within the
pharmaceutical industry and experience programming with SAS for statistical
analysis.

We seek a professional with a Master's Degree or PhD in statistics,
biostatistics, or a closely related field.

Candidates should have a sound understanding of Good Statistical Practices and
various regulatory requirements (FDA, EU, ICH), knowledge of clinical trial
design, FDA regulations, and quality standards.


Compensation:
A compensation package will be designed to attract outstanding talent and will
include a base salary plus benefits.

Not a direct answer to your question,
but in finance it is common to use a
factor model to create a variance matrix
when there are more variables than
observations.

R code to do that (including the possibility
of missing values) is available on the
Public Domain page of www.burns-stat.com.


Patrick Burns
patrick@...
+44 (0)20 8525 0696
http://www.burns-stat.com
(home of "The R Inferno" and "A Guide for the Unwilling S User")


time pass wrote:
>
>
>
> Hi all,
>
> I am dealing with a problem where I got a sample covariance matrix
> singular since it is of high dimension. Now I came across some paper
> where I got some suggestions to use ridge parameter but there I could
> not find any algorithm for finding the best ridge parameter. I am
> wondering if any one have ever faced this kind of problem could you
> please help me? Also if you have done such work do you any code for it?
> I will really appreciate your help. Thanks in advance.
>
> Thanks and Regards,
>
> Satyaki
>
>

Hi all,

I am dealing with a problem where I got a sample covariance matrix singular
since it is of high dimension. Now I came across some paper where I got some
suggestions to use ridge parameter but there I could not find any algorithm for
finding the best ridge parameter. I am wondering if any one have ever faced this
kind of problem could you please help me? Also if you have done such work do you
any code for it? I will really appreciate your help. Thanks in advance.

Thanks and Regards,

Satyaki

Hi

There is an opening for the position of Analytics Manager with a leading
Consulting Company.

Position: Analytics Manager
Location: Bangalore/Gurgaon

Job Profile:

Help the clients create valuable insights from data that lower marketing costs
and improve marketing effectiveness and efficiency.

Help the clients design/ implement marketing programs and campaigns that
stimulate incremental sales, generate repeat purchases, reduce costs, strengthen
customer relations, and extend customer life span.

Help develop and interpret statistical models for targeted acquisition,
retention, cross-sell, customer lifetime value or behavioral segmentation models
using predictive modeling techniques that include linear and logistic
regression, cluster analysis, neural nets and other statistical techniques.


Skills Required:

Experienced professional with a minimum of 6 - 12 years of relevant experience
in client management, Data Analytics, Reporting, Modeling and Regression

Education
B.Tech/MBA/Master degree in Statistics/Econometrics/ Economics

If interested, please send your updated CV in word format along with following
details -

Total Exp (Post Qualification)
Current CTC
Expected CTC
Notice Period
Current Location

Regards
Aakriti Shah
HR Consultant
Empyrean Partners
aakriti.shah@...

these are some iran pictures that are "clear evidence"

if these aren't enough please take a tour over BBC, CNN,...

all of these "clear evidence" isn't 5% of press on people

sorry, sorry, sorry

these pix are "clear evidence".

if these aren't enogh, take a tour over BBC, CNN, ...

all of these "clear evidence" shows 5% of actually news.

please help man.

hello all friends

i have master degree in statistics and i live in iran-tehran.

4 day ago we have a president election in iran. last president (ahmadi nejad)
,as a new candidate says many lies with statistics.

and after election the secret army kills and disturb many peoples in citys.

i and all of peaples are vert depress an future is unknown.

please help iran peaples with emailing this text to all of your friends.
please help.

best regards