IRAOI JOURNAL OF STATISTICAL SCIENCES

Degree of difficulty is an important concept in the classical
Geometric Programming (GP) theory. The dual problem is often
infeasible when the degree of difficulty is negative and very little
subjects have been published on this topic. This paper presents the
basic concepts and principles of multiple-objective geometric
programming model, and developed a numerical procedure to solve
multi-objective Geometric Programming Problems (GPP) having a
negative degree of difficulty using weighted method to obtain the
non-inferior solution and using Brickers simple technique to ensure
the dual feasibility; namely the addition of a constant term to the
primal objective function

Models play a critical role in statistical data analysis. Once a
model has been identified, various forms of inferences such as
prediction, control and information extraction can be
established, there is no doubt the information criterion play a great
role to identify the correct model. One of the known information
criterions is the SIC , within this research a study of the SIC
criterion behavior and it's empirical power to identify the correct
model out of many adapted different samples with sample sizes
(30,50,100) and with wide range of allowed parameter values for
models, Hence the generated time series data are from AR(1)
,MA(1) ,AR(2) and MA(2) to discover the influence factors that
affect the accuracy of this criterion .Since the criterion would be
considered the best trusted choice for identifying the correct model
in time series.
Searching for the weak points of this criterion overcome what that
is beyond it, regarded as an important stage to reach reliable
results.
To reach this target simulation techniques was used in
studying the behavior of SIC criterion
The outcome results from models parameters values effects
SIC criterion performance. Hence it was identified. Hence wrong
models were identified, which researcher most probable would
choose and fall for it.

In this paper, we have investigated two new conjugacy coefficients
for Conjugate Gradient (CG) methods for solving nonlinear
unconstrained optimization. First six-parameter family which depends
on six parameters (λ,µ,ω,δ ,γ ,ϕ ), second seven-parameter family
which depends on seven parameters (λ, ). µ,ω,δ ,γ ,ϕ,ψ The clear
feature of these two families is obtaining eleven or twelve different
forms of original conjugacy coefficient from six and seven-parameter
family, respectively. The global convergence analysis of these two
families had been proved. Numerical results for these two families on
some well-known test functions were reported and compared with the
result of some well known formula for β k in CG-methods.

In this paper, we use algorithm models of Hidden Markov models,
which concluded the possibility of recognizing the normal face as well as
the face with boundary brims by using (Canny) method. By using other methods, boundary brims of the images can be recognized using (
Sobel, Log) which gives good rate of recognition, but Canny method was
the
best way. Programs of all algorithms were written using Matlab.
(R2008a) 7.6.0

In this paper, we used the Algorithm of Hidden Markov Models, to
know gesture of hand as a frame taken by video post with three
movements and we use Threshold Model within algorithm. We suggest two threshold model and we find that Hidden Markov Models
used with a threshold model give good results with best smallest
replication. A all algorithms are writhen using the language Matlab.
(R 2009a) 7.8.0

The research deals with the employing of the factor analysis
approach to structure the transfer function models and then predict
depending on the principle component method and then apply that on
the climate data (temperature, wind speed, solar radiation and the
relative humidity) related to the evaporation - transpiration
phenomenon, considering that the climate data is the input for the
model and the evaporation - transpiration phenomenon stands for the
outputs. The model was estimated twice: in the first estimation the
transfer function was estimated using the data resulted from the factorial analysis as inputs, while in the second estimation, the
transfer function model was estimated using the climate data as inputs
and then determining the best approach in prediction by means of
prediction accuracy criterions.

Chaotic phenomena are getting interest in all spheres of knowledge
. In this paper, we use a new binary technical test, the 0-1 test for
chaos , due to Gottwald & Melborne and adopt an effective tool in
testing the chaotic dynamic systems for both discrete dynamical system and continuous dynamical system . In the past there were
certain tools to identify regular and chaos in deterministic dynamical
system such as time series curves , phase plots , bifurcation diagrams ,
lyapunov exponent etc. these tools though very powerful , are not
sufficient to differentiate regular and chaos SYSTEMS . In this paper,
the binary test has been discussed and its application has been shown
with satisfactory results . The Tent map, Tinkerbell map, Henon map,
Lorenz system have been studied and their behavior has been
observed with respect to these tools and a comparison of binary test
with these tools has been done .It has been found that the results
obtained by the binary test are quite satisfactory and significant .

The main objective of pattern recognition methods is to
develop capacity in the simulation of the most special objects in a
different form, in order to classify these objects to a number of classes
or types, these classes can be images or data or any kind of standards
that need to be rated. In this research, recruitment of and the nearest
neighbor method to the decision-making process in non-parametric
pattern recognition through the use of theory in the Bayse theorem as
a decision rule of discrimination, and their application in the field of
groundwater quality in the province of Nineveh. Has given way
suggest accurate results in the process of classifying data into two
groups: the first water wells for drinking and other non-potable water,
which could study the characteristics and qualities of each group and
to identify qualities essential so it is easy to distinguish samples from
any well in the province to see the quality of groundwater for The well
under study

The subject of geometric stochastic process is regarded as one
of the subjects that currently meet great and wide concern among
researchers currently due to its importance in different fields. The
geometric process is regarded as a generalization to the renewal
process.
This research is devoted to the study and analysis of the
geometric process, and the comparison between the geometric process
model and the renewal process model, as well as the estimation of the
geometric function through the use of the numerical solution and the
approximate solution. It includes actual application about the cases of
infection with viral hepatitis types A, B in Nineveh governorate. A
statistical analysis is performed for the geometric process, as well as
the test for suitability of the data to the geometric process, the
expected number of infection cases with viral hepatitis type A and B,
was estimated with comparison between the methods estimation.
This research revealed that the geometric process is suitable for
the data of viral hepatitis, and that the geometric process model is
better than the renewal process model. The study also showed that the
expected number for cases of infection with viral hepatitis by using
the numerical solution is better and more accurate than using the
approximate solution in estimation.

The process of classification of data usually put conditions or
limitations which are often illogical and incorrect,so in this research is
a classification process has been conducted depending on the style of
artificial neural network , specially kohonen network , which also
called self organizing maps.
The results of classification are blind without interference from the
researcher in the development of any restriction or condition and this is what is characterized kohonen network being one of the networks
that train themselves without any superviision. Valuable Results were
obtained in good appropriate time and the results of this method is to
build Kohonen algorithm for the classification of years depending on
the level of rain, and it is striking to obtain the results of classification
using the above algorithm in a very short time.

In this study, the use of cluster analysis to classify
(24) wells in the aquifer of water existed within the Pilaspi
Formation in the area of Bashiqa region depending on the quality of
ground water in order to determine whether water quality samples in
different locations could be joined to homogeneous areas. The wells
locations for the groundwater on the topographic map of the study
area have been determined according to the latitude and longitude
coordinates by using Global Mapper v.7.It has been observed from
the data obtained on the wells that it they contain six
attributes (variables) in different measured units. Data were analyzed
by using the software Minitab v.14 in a multivariate technique that
used cluster analysis. The gained results could be considered
positive. The wells were divided into groups based on the Electric
Conductivity (E.C) to know water competence range human,
agricultural, animal and industrial use graded according to portability
of water.

In this paper, we use a hybrid technique from artificial neural
network and linear regression model. We explain how this hybrid
technique works and compare it with neural network and linear
regression model. To show how well done, we used the Mean Square Error (MSE) criteria through using simulation and real examples, we
show that this hybrid technique, generalized regression neural network
has minimum value of MSE.

In this research we used Artificial Neural Networks approach to
estimate parameters of multiple regression linear model, we put
algorithm to estimate the parameters of linear and non linear
regressions models, so check the efficiency of this method, which is
compared with results of the known traditional approaches, the used
least square, that an a proved high efficiency in estimation becaused it
has the least mean square error.

In this paper we used panel data models and how to choose the
fitting three panel data models, the pooled regression model, fixed
effects model, and random effects model, depending on the adjusted
coefficient of determination and Akaike information criterion in
selecting the best sub-model form the general model through
studying the four affected variables on the public budget. This study
included of eighteen Arab countries for the period from 1995 to 2009.

The research, used the genetic algorithm as one of intelligence
research techniques to find the least cost of constrained economic case
for which Lagrange method is used to solve. The researcher suggested
a new genetic algorithm to solve this type of economic cases and find
the least possible cost as well as the critical values which made this
cost the least possible one

The analysis of time series is the most commonly used statistical
methods in the areas of daily life where phenomena are analyzed for a
certain period of time and then to predict the future In this research,
this analysis is based on factor analysis, which is a platform for
statistical analysis of multiple variables associated with each other
with varying degrees of association through the development of
these variables in the form of a list of independent ratings on the basis
of the quality of classification.
This research aims to present a method for modeling time series
depending on the independent components and fuzzy logic through
(ARIMA) models appropriate for each time series, and then find the
speculation of the time series of independent components in a manner
and method of fuzzy logic and comparison between them depending
on the standard MSE.