Show: 25 50 75 100 Results

Search results: 25 out of 1,316

مقارنة بين طرائق تقدير انحدار الحرف العامة في معالجة مشكلة التعدد الخطي شبه التام مع تطبيق عملي == A Comparisons Among The Generalized Ridge Regression Methods of Estimators Under Multicollinearity Problem With A Praxis

Author name: حنين مراد يوسف الصالحي
Supervisor name: سجى محمد حسين الهاشمي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: يلقى موضوع خرق احد فروض الانحدار اهتماما واسعا وواضحا في معظم الدراسات لذلك تنبع اهمية هذا البحث من خلال الكشف عن احد هذه الفروض ومعالجتها والمتمثلة بـ وجود علاقة خطية بين اثنين او اكثر من المتغيرات التوضيحية والتي تدعى بمشكلة التعدد الخطي (Multicollinea | The violation of regression assumptions is one of the interesting topics in many fields. This thesis deals with multicollinearity problem by using some of generalized ridge regression methods (GRR, MJR, GJR, GL, AUGRR, and AUGL) as well as our proposed method (Almost unbiased Generalized Jackknife Ridge) (AUGJR). In addition, these methods are compared with parameter (k) in ridge and parameter (Liu D). These methods are evaluated based on the mean squared error (MSE) to obtain the best method of these methods with the suitable parameter.To evaluate these methods, we use simulation studies by using the R statistical program. Five variables are simulated with different sample sizes (n=15,20,60,300), different variances (0.16,0.13,0.18) and different correlation coefficients (r=0.16,0.13 and 0.18) with (s=10 and 2). The results show that AUGL, AUGJR and AUGRR are the best methods in terms of the MSE. The differences are very small among them. The methods are also evaluated based on real data example (Rail transport for the passengers and cargo), which is obtained from Iraqi Ministry of Planning.We first detected the problem of multicollinearity by using the Variance Inflation Factor (VIF) and Condition Index (CI). Then, we build a model of revenue for transportation after they have been estimating its features at the best methods.

افضل طرائق المعاينة في تقدير اعداد النخيل في محافظة البصرة == Optimum Sampling To Estimate Date Palm In Basrah Governorate

Author name: حنان سعيد مالك الراوي
Supervisor name: رفعت لازم مشعل الخميسي | عدنان شهاب حمد
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: مما لاشك فيه ان لكل اسلوب من اساليب المعاينة مزايا ومبررات في استخدامه، وغالبا ماترتبط عملية اختيار اسلوب المعاينة بجانبين اساسيين، اولهما معيار التجانس بين المفردات سواء على المستوى الكلي ام على مستوى اجزاء(او مجاميع) من المجتمع، وثانيهما الامكانات المتا | There is no doubt that each technique of sampling techniques has advantages and reason for using it, and mostly the correlation of choosing sampling technique with a essential two sides, first of them is the standard of homogenous between units as for all population, and the second one is the abilities that are available to fulfills survey. And in using Stratified Random Sampling Technique in evaluating the suitable population to apply an important role to get a high efficiency estimator to compare with other Sampling design. As the target of this searching is to get the best ways that increase the accurate estimation of the Date Palms number in Basrah governorate and in using different kind of sampling estimator, the comparison has been done between the mean variance of Simple Random Sampling and the mean variance of the Stratified Random Sampling, and it is clear in the comparative results hat using Stratified Random Sampling give the best estimation. The determination of strata boundaries (strata formation) have interested role in obtaining efficient estimators, several strategies of strata formation were constructed, it is proved that the approximated frequency strategy at ?= ½ is the most efficient, researchers discussed developing this strategy through using different values of (?), or different sample allocations. In this searching three approximately frequency had been applied , and to divide the population into strata and point out the optimum limit strata, and the sample allocation has been determined by using Neyman distribution. And it is clear in the result that using the approximated frequency is the nearest and efficient to estimate the Date Palms number in Basrah governorate, by counting the Stratification mean variance and the stratification mean for the three approximated frequency. And the reason to know wither the Stratified Random Sampling is efficient and better than the Simple Random Sampling in evaluating the Date Palms number the experiment of the hypothesis between two population has been used, the results of that experiment show the difference between the two means that lead to reject the Null hypothesis and to accept the Alternative hypothesis, according to this the search recommended to use the Stratified Random Sampling according to the new divided stratification by using frequency, taking into consideration using Neyman distribution o determine sample size and sample allocation

تقدير الدالة اللامعلمية للبيانات العنقودية == Nonparametric Regression Function Estimation of Clustered Data

Author name: حلا كاظم عبيد
Supervisor name: سجى محمد حسين الهاشمي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: البيانات العنقودية تظهر في الكثير من العلوم الاجتماعية والصحية والسلوكية. وتتميز هذا النوع من البيانات بوجود الارتباط بين مشاهداتها. وممكن التعبير عن العنقدة من حيث العلاقة بين القياسات على الوحدات ضمن نفس المجموعة فان النماذج الاحصائية تحتم على حساب الار | Cluster data appears in a lot of social, health and behavioral sciences. And featuring this type of data link between the presences of her observations. And possible expression of clustering in terms of the relationship between measurements on units within the same group, the statistical models makes it imperative for the link account at every level, because failure to do so leads to misleading results. Hence the importance inside the Observations link to the estimating of the function non parametric for cluster data where the use of parametric method for ICON is always desirable to estimate some functions Because of the shape of the data is unknown in advance the appropriate function or as a result of the existence of some obstacles so it is the use non parametric method to estimate (smoothing) Nonparametric function.. Research has shown developed in recent times on the use of non parametric regression when parametric the assumptions are unfulfilled. And non parametric regression allows greater flexibility of functions dependent variables resulting from the data. Previous research has touched on the case of cluster data estimating the ways non parametric and semi parametric methods and was adopted state of neglect of the link within the same cluster property data that distinguish cluster data is particularly. And local kernel estimator achieved more efficient negligently correlation within clusters (even if the correlation is in the interest the study). While some touched on the case taking correlation between Observations per cluster using the estimated equations. Others had created the kernel methods in the case of cluster data behave completely different from the behavior of the capabilities of the spline estimator as has achieved kernel methods results more efficient when the neglect of the link within the clusters, while spline methods results achieved less variance of smoothing fixed parameters at taking the link inside clusters into account in the estimation process.So in this thesis will be nonparametric function estimating for clustered data using the Seemingly Unrelated Kernel Estimators, and The Generalized Least Squares Smoothing Spline Estimators and propose Robust methods and comparison of the methods listed above to indicate the best estimate of the nonparametric function estimating for clustered data, taking into account the structure of the link within the clusters were cluster data, The adoption of cluster data, which has the same number of explanatory variables within each cluster. To achieve this, thesis was divided into five chapters, the first chapter included introduction and aim of the research and reference review, either Chapter II now include the theoretical side which discussed the methods used to calculate the non parametric function of cluster data in the presence of the link. While included Chapter III experimental side (simulation) and the application addressed method in the second chapter and the statement of the best way has less (MAE) or (MSE). and either the fourth chapter includes the applied side to the real data for the proportion of white blood cells and its impact on the proportion of blood per patient (cluster) and Chapter V which includes the most important conclusions and the recommendations.it is through simulation experiments have been finding the best way to estimate the non parametric function for cluster data and a way The robust Generalized Least Squares Smoothing Spline Estimators in the case of a correlation. It was the application of all methods of the practical side using real data about the proportion of white blood cells and their impact on the proportion of blood hemoglobin for patients with blood cancer (leukemia).

التحليل الاحصائي لتجارب القياسات المكررة للبيانات المصنفة == The Statistical Analysis For Experimental The Categorical Data of Repeated Measurement

Author name: حلا كاظم عبيد الصبيحاوي
Supervisor name: سجى محمد حسين الهاشمي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: حيث تم استعمال اختبار (Cochran, Mc Nemar, Ireland & Kullback, Stuart, Bhapkar, Ireland & Ku & Kullback) على بيانات ذات معالجتين وكل معالجة بمستويين المتمثلة ببيانات مرض ضغط الدم ولدوائين وكل دواء بمستويين (ملائم وغير ملائم) ومن خلال الجانب التطبيقي تم الو | In this dissertation the study have concentrated on the use of repeated measurements tests of catcorigal data, where we used (Cochran, Mc Nemar, Ireland & Kullback, Stuart, Bhapkar, Ireland & Ku & Kullback) tests on two treatments data, and every treatments with two levels which represent raised blood pressure data, and for two medications, each medication with two levels (appropriate and inappropriate), through the practical aspect, we concluded that the ratio of response for the two medication are the same. Also we used (Ireland & Ku & Kullback, Bhapkar, Stuart) tests on data with two treatments, each treatment has more than two levels that represent data of fifth grade (High school) for Arabic text book through years (2001 - 2004) in sumer high school, in which the first treatment represents the students grades in mid term exam, and the second treatment represented the student grades in final exam, through the practical aspect we gained an equivalent results in the previously mentioned methods (in which the students levels in mid term exam equivalent to that in final exam). We also used (Ireland & Kullback, Cochran) tests and weighted least squares (WLS) tests on three treatments data in which every treatment has two levels that represents data of student in the same grade, and in the same years for English text book in which the first treatment represents the first term average, and the second treatment represents the student's grade in mid year, and the third treatment represents the average of second term. And for two levels (Pass and failed), through the applicative of these test on these data we concluded that the level of any given student in first term is equal to the level in mid year and it's also equal to that in second term.

مقارنة طرائق تقدير دالة البقاء لتوزيع لوماكس باستخدام عينات مراقبة من النوع الثاني

Author name: حلا سلمان فرحان
Supervisor name: صباح هادي عبود الجاسم
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: في هذه الرسالة تم تقدير دالة البقاء لتوزيع لوماكس اخذين بنظر الاعتبار وجود عينات مراقبة((Censored Data من النوع الثاني في حالة توافر معلومات اولية عن المعلمات بشكل دالة احتمالية اولية مشتركة لكل من معلمتي الشكل والقياس,حيث تم افتراضها حسب اسلوب الباحث(Jef | In this study have been estimated the survival function to Lomax distribution considering there was censored data (second Type) just in case if a primitive information were available about the indications as mutual primitive probability function for both : indications form & measure which assumed in order to method of the searcher (Jeffry) and also squared error loss function). So measured estimated pez S*(t) for survival function S (t) is the conditional expectation function S*(t) =E[S (t)/X].It has been used approximated methods to measure them because It is hard : to measure the complementary of numerator and denominator in a mathematics form by using Bayes method. One of these methods is the method of the searcher ( Lindley) while the other method was return to ( Tierrney and Kadon).In addition to methods formerly to estimate the survival function; the searcher has used the Shrinkage method. While she used the maximum likelihood method as a usual method for those which did not depend on the primitive information because the estimated thing had fixed quality.So, all this to make the comparison between these methods and the oldest one through simulation style by Mont - Carlo to get new estimation which carried the wanted qualities in the perfect estimated to get specific results : to choose the suitable estimation method.The searcher hade made the bayes (lomax distribution survival function) as the lindley manner was the best among the other manners to estimate lomax distribution survival function ,on general ;while the manner of the maximum likelihood method was the best manner among to other estimated manners for tiny simple size on private.

مقارنة فترة الثقة مع الفترة البيزية للتركيبة الخطية لمتوسطات عامل التداخل مع تـطبيق عملي == A Comparison of Confidence Interval With Bayesian Interval For The Linear Compination of Nest Factor Means (With Application)

Author name: حسام موفق صبري الدليمي
Supervisor name: اموري هادي كاظم الحسناوي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: تعد دراسات المقارنة بين طرق تقدير المعلمات من الدراسات المهمة التي تاخذ حيزا واسعا في البحوث الاحصائية، لذلك تهدف هذه الدراسة الى اجراء مقارنة من خلال الفترات بين فترة الثقة والفترة البيزية للتركيبة الخطية لمتوسطات عامل التداخل في تصميم متداخل متزن لمرح | A Comparison studies among parameters estimation approaches are important studies that take a large space in statistical searches. So that the purpose of this study is to make a comparison through the intervals between the confidence interval and the Bayesian interval for linear combination of nest factor means in a two stage balanced nested design.The confidence interval was found by using an advance approach in inference called (Mixed Inference), and the Bayesian interval was found by using another approach in inference called (Bayesian Inference), then a comparison was happened between the two intervals (obviously the two kinds of inference).We apply this comparison on a real data experiment represent weights of vetch planet in (1998), and by using simulation. The results of the real data experiment and simulation shows an important conclusion which said that the confidence interval for linear combination of nest factor means is better than the Bayesian interval of the same linear combination for any suggestion confidence coefficient.Finally, An assumption study established to know the causes that leads to that important conclusion

تحليل التباين متعدد المتغيرات لتصميم القطع المنشقة - المنشقة == Multivariate Analysis of Varaince For Split - Split Plot Design

Author name: حسام عبد الرزاق رشيد البكري
Supervisor name: ظافر حسين رشيد النجار
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: This research work is a study of multivariate analysis of variance model (MANOVA) for the design of split - split plot to a complete randomized block. As it is an important method for determining whether the response variables are influenced with each other or in other words whether the correlation between the variables of a force such that it influence the decision taken concerning the applied hypothesis. When we don’t ignore the relationships which are created between the responses variables lead us to generalize the tests in the univariate analyses about the means of normal population to be changed to tests involving mean vectors responses taken from multivariate normal population. Ignoring the existing relationships between variables lead us to untrue conclusions, so in doing the (MANOVA) gives us a full picture of the experimental factors and their interaction in the experiment as a whole. The concentration on doing the (MANOVA) should not be at the expense of the study of the univariate analysis. So the research work involves the problem of heterogeneous in sub - sub plot variances, which is considered one of the important problems put forward by other researchers which may limit his work in special procedure. We are interested in solving this problem by using a univariate analysis procedure after splitting the error of sub - sub plot and then finding the standard deviation used in comparison testing between about the third factor (c). Then comparing them with the standard deviation used, if we assume the existence of homogeneity in variances of sub - sub plot.

استخدام اسلوب بيز التجريبي في تقدير معلمات انموذج الانحدار الخطي == Using Empirical Bayes Approach For Estimating Parameters In A Linear Regression Model

Author name: حازم منصور كوركيس عربو
Supervisor name: اموري هادي كاظم الحسناوي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: The primary purpose of our thesis is to assess the importance of using empirical Bayes approach in regression analysis. At the first some empirical Bayes techniques in point estimation are considered. It is well known that in point estimation with a squared error loss function the Bayes estimator is the posterior mean. In the empirical Bayes approach we must construct a consistent sequence of estimators for this posterior mean using past experience. This construction is done for three general families of distributions. New estimators for the parameters in the simple orthogonal linear regression model are presented using not only the usual random sample of observations but also past experience in the form of previous estimators of parameters in similar but independent situations. The regression parameters are considered to be random variables. Bayes estimators are given for a squared error loss function. Even though the prior density of the parameter is unknown the Bayes estimator can be written in terms of the marginal density of sufficient statistic. This marginal density can be estimated empirically , thus forming the empirical Bayes estimator. Empirical Bayes estimators for the parameters in the general linear regression model are presented. These estimators by pass exact knowledge of the prior distribution of parameters by means of supplementary informations from similar independent experiments. The case in which the error variance is unknown and may vary from one experiment to the next is included. Some basic concepts in shrinkage estimators are introduced. The definition of multicolinearity as the existence of near linear relationships among the independent variables is given. Effects of multicolinearity on estimated regression coefficients are explained. Sources of multicolinearity and methods of detecting multicolinerity are presented. The method of ridge regression is given as one of several methods that have been proposed to remedy multicolinearity problems by modifying the method of least squares to allow biased estimators of the regression coefficients. The technique of ridge regression first proposed by Hoerl and Kennard has become a popular tool for data analysts faced with a high degree of multicolinearity in their data. Ridge solution properties and methods for choosing the ridge parameter are presented. The first method is graphical applied by using graphical display called 'ridge trace' the second method is the iterative method proposed by Hoerl and Kennard. The equivalence of ridge regression estimator with Bayers estimator is proved. Bayesian methods are employed for choosing the ridge parameter. An empirical Bayes estimator of the ridge parameter is presented. An empirical Bayes estimator of the ridge parameter which result in minimax ridge regression estimator under strawderman's loss function (formula 2.81) is also presented. By minimax estimator we mean an estimator which is uniformly better than the least square estimator in terms of risk. In the practical part of the thesis we apply ridge regression analysis to the set of actual data suffer from multicolinearity. Three methods are employed to determine the value of the ridge parameter. Comparisons between the three methods are made on the basis of various statistics that might go into the choice of the ridge parameter. According to these comparisons we conclude that the value of the ridge parameter obtained by using empirical Bayes approach (formula 2.78) is better than the other two methods. Ridge analysis is repeated for another set of experimental data obtained by constraint simulation. The same conclusion is obtained when the dependent variable is any one of the variables Y2 ,..... Y6.

مقارنة بعض طرائق التقدير اللا معلمية لنموذج الانحدار التجميعي المجزا باستعمال المحاكاة مع التطبيق == A Comparison of Some Nonparametric Estimation Methods of An Additive Quantile Regression Model Using Simulation With Application

Author name: جنان عبد الله عنبر
Supervisor name: دجلة ابراهيم مهدي العزاوي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: ان استعمال النماذج المعلمية يتطلب توافر معلومات كافية عن الظاهرة المدروسة مع معرفة المجتمع الذي سحبت منه العينة وان تكون معلماته معرفة لكي تكون قراءة هذه النماذج قراءة صحيحة كما تتطلب وجود بيانات من النوع الكمي الامر الذي دفع الباحثين الى البحث عن نماذج ا | Using of the parametric models requires a number of preliminary conditions that should be available to make the reading of these models right as well as the presence of quantitative data which motivated the researchers to search for models with lesser terms than the parametric models represented by nonparametric models, so too many considerations have been made for the Nonparametric Regression, in the last years. The main reason of that was the researchers’ belief that pure parametric methods for estimating the regression curve have shortage with the flexibility needed for data analysis - especially the quantitative data - . With the progress made in computers, in both hardware and software, it has become possible to develop many of Nonparametric Regression Methods including the "Quantile Methods". In spite of that, the researcher see the most of papers and articles concern univariate case, where the researchers about bivariate case still with limited. In the other hand, the researches extension to multivariate case nonexistent in Iraq - in the scope of researcher’s knowledge - , which gives an extremely importance for the aim of this study, to bring attention for alternative methods or modified methods that can be efficient ones to improve the currently methods. Thus, the most important purpose of the research, is the use of Tow - Stages method, which helps the researchers to compute Nonparametric estimators for the components of Nonparametric models to avoid the curse of dimensionality.It divideds the components to set of points called "Quantiles" , then estimates the quantile function by one of the quantile methods we explained in this research, ( Marginal Integration, Backfitting, and Tow - Stages ). The simulation has been designed for the illustrated models and it has been checked about the estimation methods performance by using the Absolute Deviation Error (ADE) criterion, then compute the Average Absolute Deviation Error (AADE). From noticing the simulation results, it was shown that the best estimator was Tow - stages estimator in case of high correlation and / or high dimensions. To achieve the purpose of this research, the study has been divided into five chapters. Chapter one, consists of an introduction, the aim of the thesis under research, and a literature survey. Chapter two covers the parametric regression quantile estimation. While Chapter three devoted for Nonparametric estimations methods. Chapter four devoted for Simulation experiments and practical experiment with real data. Finally, Chapter five comprises the conclusions and suggestions that the research has recommended.

مقارنة بعض المقدرات البيزية الحصينة مع مقدرات اخرى لانموذج GARCH(1.1) مع تطبيق عملي == A Comparing of Some Robust Bayesian Estimators With Another Estimators For Garch (1.1) With Practical Application

Author name: جنان عبد الله عنبر
Supervisor name: نزار مصطفى جواد الصراف
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: تعاني بعض السلاسل الزمنية من التقلبات او عدم الثبات في التباين مثل السلاسل المالية والاقتصادية والبيئية وغيرها, وقد يرافق ذلك وجود التلويث او القيم الشاذة في تلك السلاسل والذي يرافق عملية جمع البيانات في اغلب الاحيان ولاسباب عديدة قد يؤثر ذلك بشكل كبير عل | Some of time series suffer from volatility or instability in variation, such as financial , economic , environmental and other time seriesIt was accompanied by the presence of contamination or stray values in those chains that accompanies the data collection process often for many reasons, which greatly affect the estimation models parameters and thus makes the estimated models parameters and thus makes the estimated models are inaccurate and affect the future in the forecasting process this makes the process of estimation the traditional methods is not accurate and not feasible in practice and that is what led many researchers to find alternative methods of estimating for those methods reduce the impact of contamination and the volatility in the process of estimating the time series models,, including autoregressive conditional heteroscadestic models family (ARCH and GARCH). So the goal came thesis complement the work of researchers as thesis aims to find robust Bayesian estimators to the estimate first order generalized autoregressive conditional heteroscadestic model GARCH (1.1) when errors followed normal distribution, and that by proposing three robust Bayesian methods to estimate a method (y ?BM.Bayes) and method (BM.Bayes) and the reduced method (BM.Bayes Shrinkag). As was the use of certain methods of estimation models (GARCH), such as (MLE) traditional method of estimation and the method of (Bayes) and three robust bounded methods a (BM.Huber) and two methods by the proposed (BM.Hample) and (BM.Tukey). The use of simulation in the style of the experimental side for a comparison between the methods adopted in research using polluting ratios (0% 0.1% 0.10% 0.15% 0.20%) and volumes of samples (500, 1000.1500), In addition to the use of different values of the parameters it is found favorable proposed method (BM.Bayes) be when the values of the two parameters (?1, ?) close to each other when any correlation strength is high , Simulations were also on the values of the parameters of the real series that have been estimated in a manner program application (MLE) and some of them were far from any values that weak correlation strength , It turns out that the best method was the proposed (BM.Bayes.Shrinkag). In the practical side it has been stated in the application of the theoretical side of the building stages of the model and testing of those stages on a series of (1254) Show prices daily sales of Basrah, for the period (2 \ 1 \ 2008 - 31 \ 12 \ 2012) through the application of the proposed third method (the reduced method) (BM.Bayes Shrink) which was best when applied to the estimated values of the parameters in a manner (MLE) in the experimental side as it made less (MSE) and estimate the appropriate model GARCH (1,1) proposed the adoption of the reduced way (BM.Bayes.Shrinkag).

استعمال بعض النماذج الاحتمالية المنفردة والمركبة المبتورة لتحديد خصائص التعويضات الصحية في شركة التامين العراقية == Use Some Probability Single & Compound Truncated Models To Determining The Characteristics of Health Payments In The Iraqi Insurance Company

Author name: ثائرة نجم عبد الله الامير
Supervisor name: قيس سبع خماس
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: نظرا لعدم وجود بحث احصائي سابق لسلوك تعويضات التامين الصحي الذي يمثل النسبة الاكبر من التعويضات في عموم شركات التامين في العراق، تم اختيار هذا البحث وتطبيقه في شركة التامين العراقية.ولغرض تحديد افضل انموذج احتمالي يمثل تعويضات التامين الصحي، تم استعمال | Due to the lack of previous statistical research of the behavior of payments, specifically health insurance, which represents the largest proportion of payments in the general insurance companies in Iraq, this study were selected and applied in the Iraqi insurance company.To determine the best model represent the health insurance payments, we used two probability models determined through the initial detection for the distribution of the research sample by use (Easy Fit) program.One : single, a (Lognormal) for all sample views, and the other compound (Compound Weibull) at dividing research sample into small losses and large losses, and focused on the compound model in some detail in terms of drafting and its importance. With the application of the state of amputation in both the fact that the specific health payments from the top by two million diners in this company.Both models Parameters were estimated using the maximum likelihood method (MLE) and the use of style (Newton - Raphson) to find these estimates. And then compare between models using standard (MSE). Was reached in general that the compound model is better than a single model in the representation of payments.

طرائق تقدير عدد مرات الفشل في الانظمة القابلة للاصلاح

Author name: ثائر فيصل شاهر
Supervisor name: عبد المجيد حمزة الناصر
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: ان الهدف من هذا البحث تسليط الضوء على احد اهم وابرز المواضيع المعاصرة في دراسات المعولية وهو الانظمة القابلة للاصلاح والذي يعني ان النظام الذي عندما يحدث فيه فشل فيمكن اعادته للعمل باصلاح بعض مركباته دون الحاجة الى ابدالها. والذي يحظى بتطبيقات واسعة وخاص | The main purpose of this research was to study one of the main modern subject which is very important in the reliability studies, it is the repairable system, which means that if the component fails it immediately repaired.These studies have widely applications in many systems like watching machine of cars, airplanes and communication systems, that is failure make huge materiel and humane losses.It is found that these systems submit to Poisson process in particular the nonhomogenous Poisson process, the main contribution in this research is the modification in estimating the number of failures in repairable systems, and the derivation the distribution of n - failures where the distribution is General Gamma.Finally, some new results obtain and a simulation experiments were done to compare the proposed and classical methods.

مقارنة مقدرات بيز الحصين مع مقدرات اخرى لتقدير دالة المعولية التقريبية لتوزيع ويبل == A Comparison of Robust Bayesian Estimators With Another Estimators To Estimate The Approximation of Reliability Function For Weibull Distribution

Author name: تهاني مهدي عباس الياسري
Supervisor name: مهدي محسن اسماعيل العلاق
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: من الملاحظ ان معظم البحوث والدراسات في مجال الاحصاء ولاسيما في حقل المعولية تهدف الى الحصول على مقدرات ذات مستوى عال من الكفاءة. لاسيما عندما تكون بيانات العينة قيد الدراسة ملوثة. او في حالة عدم التحديد الدقيق للمعلومات المسبقة او في حالة ظهور الحالتين م | It is noticed that most of the researches and studies in the field of statistics particularly in the field of reliability aiming to obtain estimators of very high standard of competency specially when the sample data under study is contaminated, or in the case of not determining accurately the previous information or in case both cases appear together. From here came the objective of the thesis to come to competent estimators of the approximate reliability of Weibull distribution which is considered one of the common failure models. Through studying the ordinary procedures represented in the maximum likelihood method and method of moments and white’s method as well as the ordinary robust methods represented by the method of M - estimate and bayes procedure as well. Therefore bayes procedure was suggested depending on natural conjugate prior function in addition to suggesting three methods according to robust bayes procedure where the first suggested procedure of robust bayes assumed that there is an extent of belief and to a certain percentage with the researcher (the percentage of the researcher falling into mistake in determining the primary information) where the accurate determination of the primary information has already been made which will be integrated with the sample prospects to obtain bayes estimater As to the second suggested procedure of the robust bayes it treated the problem of the sample prospects contamination or the primary information. Whereas the third robust bayes procedure suggestion had treated the problem of appearance of both problems together - that is inaccuracy in determining the primary information and the possibility of contamination to the sample views or the primary views as well, found a comparison between the studied evaluation methods by using the simulation procedure. It has been arrived to the success of the suggested methods in evaluating the approximate reliability of Weibull distribution depending on the two statistical scales (Integral Mean Square Error) (IMSE) and the (Integral Mean Absolute Percentage Error) (IMAPE).

نظام خبير محوسب لتحديد الفترة المثلى لعمليات الصيانة == Computerized Expert System To Determine Optimal Interval For The Maintenance Process

Author name: بهاء عبد الرزاق قاسم
Supervisor name: محمد عبود طاهر
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University Of Basrah - Faculty Of Administration And Economics
Language: Arabic
University location: Basrah
First pages:
Abstract: Maintenance process is considered to be one of the important activities in the life of industrial establishments, thus the maintenance aim to conservatism on high operational properties of machines.Depending of precedes, we suggested a system instead of a statistical expert in the ( Domain of determination of optimal interval for execution maintenance policies ).The System has a statistical knowledge base about, weibull and exponational distribution, in addition to a knowledge about test of goodness of fit to determine the probability distribution of delay time.After we build expert system , we test this system as first phase on the data of past study, and we get outcome accordance to study outcomes, then we supply guide on expert system success.Stat company of fertilize south region has been chosen to be a good field for this study , and we obtain to optimal interval of maintenance execution equivalent to four month.

تقدير دالة المعوليـة لتوزيع لوغارتيم الطبيعي مع تطبيق عملي == Estimate of Reliability Function For Lognormal Distribution With Practical Application

Author name: بلسم مصطفى شفيق النائب
Supervisor name: صباح هادي عبود الجاسم | فرحان اسماعيل العاني
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: لقد كان الاهتمام الواسع والمتزايد في دراسة موضوع المعولية في السنوات الاخيرة يعود الى التطور التكنولوجـي والتقني السريـع واستخدام الانظمـة الالكترونية المعقـدة في مختلف المجالات. وعليه فان دراسة موضوع المعولية والربط بين الجانبين النظري والتطبيقي امر له | The studies of the subject - matter of Reliability with connecting between theoretical and application sides , it is an important matter now a days because the technology development with using electronic systems.In this study it is used the Lognormal Distribution as a failure models when the failure rate to these models initially increases over time and then decreases approaching Zero with increasing time , and find out the estimators of reliability function by different methods.The estimates are : 1. Maximum Likelihood Estimation.2. Uniformly Minimum Variance Unbiased Estimator.3. Bayes Estimation.And then comparing between the priorities of the estimates through using the Monte - Carlo method and conducting several experiments using Mean Square Error.Generally result show that Maximum Likelihood Estimation to estimate Reliability is better than the other methods , Regarding the practical aspect the data of life time for four machines from EIGC and found after using chi - Square test that the life time for test machines distribute Lognormal and the estimation of Reliability function for machines done using Maximum Likelihood Method

تقدير معلمات انموذج المعادلات الهيكلية المتضمن متغيرات الوساطة مع تطبيق عملي == Estimation of Structural Equations Model Parameters With Practical Application

Author name: بشرى سعد جاسم
Supervisor name: غفران اسماعيل كمال
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: يستعمل تحليل الانحدار مع المتغيرات المصنفة الى صنفين صنف يمثل متغيرات مستقلة (Independent variables) واخر يمثل متغيرات تابعة (Dependent variables), فلذلك يقوم تحليل الانحدار بدراسة العلاقة بين المتغيرات المستقلة والمتغيرات التابعة, الا ان هذا التحليل يعمل | regression analysis use with classified variables into two class that represents the independent variables (Independent variables) and the other is a subsidiary variables (Dependent variables), for there the regression analysis study the relationship between independent and Dependent variables, but , this analysis works to know only the direct impact between the variables for this reason i use the structural equation Model (SEM) to identify and know the variables that are of indirect effects by estimating and testing parameters by set of methods (steps causal method, bootstrap method, method of multiplying the transaction ( parameters) product of coefficients, difference in coefficientsstructural equation model like other models are a matching variables tested with the phenomenon studied , test the compatibility of the variables that make up a structural equation model, and to achieve this condition, use Confirmatory Factor Analysis (CFA) way to see match variables that compose it. After confirming the conformity of the model or suitability experimenting and having the effect of mediating variable in the model and mediation are two types : Single mediation where transmission of the influence of the independent variable to the dependent variable through the mediation of a single variable, and multiple mediation where is transition Effect independent variable x to the variable y through several mediation variables. the practical side of study include the effect of cultural stat of the man (X) in the use of violence against women (Y) through a series of mediation M_1variables represent (women's empowerment) and M_2represents (family planning) and the study data are taken from the integrated survey of social and health state for Iraqi women (I - WISH) for the year 2011 in the Ministry of planning - Central Statistics organazation, and this data applied conditions of adequate to structural equation model SEM and, and then estimate the parameters mediating variables and test their ability to move the indirect effect by the methods mentioned above using a program.AMOS V.23The researcher concluded that a moral mediation variables tested when using standard errors formulas for (Sobel and Goodman and Aroian) and compensated for in the test version of z all results be close itself in the other the researcher contrast were recommendations of the research is to use a single version of the standard errors formats (Sobel and Goodman and Aroian) to test the effect of mediating variables in the model, as the researcher found that the independent variable X (cultural condition of the man) affects the Y variable (violence against women) indirectly through mediation M_2variable (family) organization.

التنبؤ باستعمال نماذج الانحدار الذاتي العامة المشروطة بعدم تجانس التباين (GARCH) الموسمية مع تطبيق عملي == Forecasting The Use of Generalized Autoregressive Conditional Heteroscedastic Models (GARCH) Seasonality With Practical Application

Author name: بريدة برهان كاظم
Supervisor name: فارس طاھر حسن الكواز
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: مما لا شك فيه، تحظى نماذج GARCH) ) بالفاعلية والشعبية الكبيرة في نمذجة البيانات الاقتصادية والمالية، اذ تسمح للتباين المشروط بالتغير عبر الزمن، مما يجعلها اكثر واقعية في المجال الاقتصادي. وتتوفر ميزة اخرى مهمة في عالم الاقتصاد، ممثلة بالموسمية، الت | Un doubtedly , The GARCH model is very popular and effectiveness in economic and financial data , since it allows the conditional variance to vary over time , which makes them more realistic for the economic world. And there is another important characteristic in the economic world , Represented seasonality , that exist in high frequency data such as daily series , it can be seen in the real data of the exchange rate IQD/USD , Because there are seasonal conditional heteroscedasticity clearly shows in this data , Thereby are dealt with this type of data using Multiplicative seasonal generalized autoregressive conditional heteroscedastic models , Because it is proven effective to express their seasonal phenomenon on the contrary GARCH models which do not contain seasonal vehicle. hence the aim of the research reaching a better model represents the seasonal data with proof of the effectiveness of the seasonal model in preference to the usual model. it has been used to detect seasonal presence in the data first , after that was diagnosed a problem of heteroscedasticity passing through the phase estimation using the conditional maximum likelihood and assuming normal distribution of errors , then determine the appropriate rank of the model using a number of special criterian Represented each of the Akaike Information Criterion (AIC), Schwartz Information Criterion (SIC) , Hannan Quinn Information Criterion (H - Q), down to the stage to predict , using two method to predict the first is the prediction in the sample , which objective was to infer the efficiency of the preferred model and the second way forecasting out of sample any prediction of future values.it is found through the application on the study data stages that the best model for predicting volatility is SGARCH (1,0)(1,0).

التحليل البيزي لنماذج الانحدار الخاصة بالبيانات المزدوجة (panel data) == Bayesian Analysis For Regression Panel Data Models

Author name: باسم شليبه مسلم عباس القيسي
Supervisor name: اموري هادي كاظم الحسناوي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: تعد البيانات الاحصائية ضرورية عند دراسة اغلب ظواهر المجتمع ومنها الظواهر الاقتصادية والاجتماعية والنفسية الخ كون تحليل تللك البيانات باستخدام الاساليب الاحصائية توفر للباحث او متخذ القرار حول الظاهرة المدروسة قاعدة من المعلومات الضرورية لتسهيل عملية اتخاذ | The statistic data is important to study most of phenomena as economical , social , psychological phenomena..etc. The analysis of this data via the statistical methods gives the researcher or the decision maker more information about the studied phenomenon to make the suitable decision. The data availability needs to limit a mathematic model which represents them by the researcher and to put in consideration the type of the available data. One of the these data is panel data which can be defined ( that they are repeated measurements to the studied phenomena for N from the cross section and for T from the time series ) which can be represented by one of the model ( fixed effect model or random effect model ).Most of the studies exposed estimation of the chosen model parameters by using the classical methods as GLS and FGLS methods , whereas this research touches on the Bayesian methods to estimate parameters , especial regressive model (panel data) ,depending on the priors distributions ( non informative prior dist. and conjunct prior dist. ) and to compare them with ML method to chose the best ones for estimation and choosing.After doing two practical applications in the research ,we obtained findings. ML method was the most significant for estimating the model parameters for the fixed effect model and pooling model. Bayes method which depends on non informative prior distribution , got the second grade.When using Bayes method depending on prior distribution used in the research to estimate the fixed model parameters when the cross sections number are little to get multivariate - t posterior distribution , and if it is used with random effect model we get multivariate normal after making asymptotic expansion with existing of a loss function of the weighed squired error.

تقدير نماذج مختلطة للبيانات المصنفة مع التطبيق العملي == Estimation Mixed Models Using Catacorical Data With Application

Author name: ايناس عبد الحافظ محمد
Supervisor name: خالد ضاري عباس الطائي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: يعتمد مفهوم التوزيعات المختلطة ( المركبة) في تطوير نماذج مناسبة للبيانات المصنفة , وقد تم في هذا البحث ايجاد بعض الانماذج الاحتمالية الملائمة لهذه البيانات ومن ثم العمل على تقدير معلمات هذه الانماذج.وبعد تشخيص اوتحديد نوعي التوزيعات وهما (التوزيع المركب | The concept of mixed distributions depends (composite) in the development of appropriate data seed models, has been in this research finding some models appropriate probability of this data and then work to estimate the parameters of this models. After diagnosis Aothdid two types of distributions, namely (compound beta - Bainomal (Beta - distribution Binomial) and the distribution of Ganerlized logarithmic Series distribution (GLSD)) has been working to estimate the parameters of these distributions methods usual such as way as possible (MLE) and the method of moments (Moment) and the method of Chi - square (Chi - squar) and methods of unconventional such as search cuckoo algorithm Hawwarzmih simulated annealing. The theoretical side included the concept of vehicle models and how to configure form by probabilistic normal function and methods different appreciation such as method Maximum Likelihood Function (MLE) and the method of moments (Mom) and a method to minimize Chi - square (Minimum) and methods of artificial intelligence techniques such as search cuckoo algorithm (Cock Search) algorithm simulated annealing (Simuannling) has been presented simulation has been adopted in the comparison between the estimation methods and we had simulated experiments at different volumes of samples( n= 20, 50, 100,250). And repeat each experiment R = 1000 to achieve the goal and were compared using statistical measurements (MSE, AMPE) found that the best method (cuck) which is proposed by the researcher. The researcher numbers Bernamjeh (Matlab, R2005 B) and put all the results in the tables either the practical side and having briefed researcher at data rates of disability and for the period of (2007 - 2010) obtained through the health center for the disabled has been the comparison between Models through criterion (AIC) the standard bayes Information (BIC) and the standard G2) the researcher found that the capabilities of the search cuckoo algorithm is better than during the experimental side, therefore this algorithm applied to real data to complement the tests of good matching, which enabled Khalalhl the researcher presented the practical application of the private tables results. It was found through statistical analysis that in the year (2011) has a lower standard Akaki for the distribution of the compound compared with the distribution chain logarithmic year, which indicates that it has the best modelFrom the conclusions that have been reached by using way (GIBS) in the simulation of mixed distribution and the method of rejection and acceptance (Reject, Accept) for distribution Genaralized logarithim Series is that the capabilities of the specimen landmarks using the best in terms possess the lowest average error boxes in sizes small samples search cuckoo algorithm is medium and large The rest of the roads were estimates varying values. It has been presented the recommendations that emerged from the thesis as well as future research

استخدام التحليل البيزي لنموذج انحدار خطي بسيط باستخدام اسلوب المحاكاة == The Bayes Analysis In The Model Simple Linear Regression Using Simulation

Author name: ايناس عبد الحافظ محمد
Supervisor name: محمد صادق عبد الرزاق الدوري
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: تبقى عملية التقدير لمعلمات انموذج الانحدار من الموضوعات المهمة على الرغم من كثرة ما كتب عنها من خلال البحوث والدراسات التي تختلف باختلاف الاساليب المتبعة في عملية التقدير سواء كان هذا الاسلوب تقليديا او بيزيا.الغرض من هذه الدراسة هو توظيف معلومات مسبقة | The process of estimating the parameters of regression, is still one of importent Subject despite of large number of papers and studies written in this subject, these studies are differ in techniques followed in the process of estimation if this techniques classic or Bayesian. The purpose of this study to prior information about the parameters was applied which estimated according to Bayesian theory which represented by Gamma distribution with one parameter in the estimating process and for varies simple sizes then making a comparison between estimation techniques represented by (OLS) method and Bayesian techniques by providing the distribution and all techniques which deals with derivatives process to obtain the Bayesian estimation. By using the simulation technique, results has been obtained to provide the researcher the image to clarify the best method in estimation technique

استخدام نظرية الرابطة لتحليل دالة البقاء ذات المتغيرين == Using The Copula Theory For Analyzing The Bivariate Survival Function

Author name: ايمان عبد علي داود
Supervisor name: ضوية سلمان حسن الجنابي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Doctorate
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: تم في هذا البحث استعمال نظرية الرابطة (Copula Theory) في نمذجة دالة البقاء ذات المتغيرين اذ تم استعمال انموذجين يخضعان لتوزيع ويبل ذي المتغيرين وانموذج اخر يخضع للتوزيع الطبيعي ذي المتغيرين القياسي المبتور عند الصفر وبنسبة خلط واعتمادية واحجام عينات مخ | It has been concluded in this research to use the copula theory in modeling the survival function of the bivariate as two models yielding to the bivariate weibull distribution has been used and another model that submit to the bivariate standard normal cut off at zero point and mixing percentage, reliance and sizes of different specimens and a simulation experiment has been carried out for the purpose of making experimental comparison between the evaluations of the survival function mentioned above by using six different copulas in addition to two common models, the first by using the function when assuming independence and the second by using the survival function definition of the bivariate as the use of copulas has been reached which led to acquire an evaluated survival function of the bivariate with properties of more fineness than the two evaluated survival functions of the bivariate which had been acquired by using the two common models referred to above.

مقارنة بين اختبار (Gold feld Quandt) الحصين مع اختبارات اخرى للكشف عن عدم تجانس التباين بوجود القيم الشاذة == A Comparison of The Test (Gold Feld Quandt) Modified With Other Tests To Detection The Presence of Heterogeneity of Variance of Outliers Values

Author name: ايلاف بهاء علوان
Supervisor name: محمود مهدي حسن البياتي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: ان مشكلة عدم تجانس التباينات في حالة وجود القيم الشاذة لها جانبان الاول : هو كيفية تعامل الاختبارات مع مشكلة عدم تجانس التباينات في حالة وجود القيم الشاذة بالنسبة لانموذج الانحدار الخطي المتعدد حيث نلاحظ ان الاختبارات الاعتيادية (الكلاسيكية ) تعاني من مشا | The problem heterogeneity in the case of the presence of outlier values has two important sides. The first is how to handle the test, which have the problem of lack of heterogeneity in the case of outlier values for the multivariate linear regression model where we notice that the usual tests (the classical) have the problems in the results, and the results obtained will be inaccurate and misleading. Therefore, these will be unreliable results, so it is necessary to use other tests to substitute the regular tests, they will work in the same way of normal teste in the absence of the problem of heterogeneity and they are called robust tests. These tests are Modified GoldfieldQuant, Modified Bayes, and Modified levene. Different percentages of data were cut which are (10% , 25% , 40%) assuming normal distribution of data. A comparison was made of the mentioned tests by using soft ware power of the test of Monte Carlo Simalation then detect the best test by force standard where Bayes robust was the best test for detecting the problem of heterogeneity in the presence of outlier valuesand gave reliable results. In the second side, Box plot was used for the detection of outlier values in real data. As for the practical side, data from the agriculture and cultivation of planning and follow - up / meteorological center were collected and used in this study on the four variables for the year 2013 - 2014 and the variables are : Raining rate (y).Air pressure (x_1).Temperature rate (x_2).Humidity rate (x_3).

الطرائق البيزية والتقليدية في تقدير معلمات بعض نماذج بواسون غير المتجانسة مع تطبيق عملي : بحث مقارن == Bayesian And Ordinary Methods For Estimating Parameters of Some Non - Homogeneous Poisson Models With Practical Application Comparative Research

Author name: ايات صادق جعفر
Supervisor name: ايمان حسن احمد
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: تعد عمليات بواسون غير المتجانسة احدى الموضوعات الاحصائية التي اصبح لها اهمية في جميع العلوم ولها تطبيقات واسعة في مختلف المجالات كنظرية صفوف الانتظار والانظمة القابلة للاصلاح وانظمة الحاسوب والاتصالات ونظرية المعولية وغيرها، كما تستعمل عمليات بواسون غير | The Non - Homogeneous Poisson process considered one of the statistical subjects which had an importance in other sciences and had a large application in different areas as the theory of waiting raws rectifiable systems, computer and communication systems and the theory of reliability and many other, also it used in modeling the phenomenon that occurred by unfixed way over time (all events that changed by time).This thesis deals with some of the basic concepts that are related to the Non - Homogeneous Poisson process, also this research mentioned two models of the Non - Homogeneous Poisson process which are the power law model , and Musa - okumto , also many different methods have been used in the estimating the parameters of the model , of which the classic methods would be used , maximum likelihood method and moment meethod to estimate the parameters of power law and Musa - Okumoto model , in addition to that the use of Bayesian method in the estimation of the parameters of the two models which are used in this research , in order to find the best way in the estimation , we referring to simulation manner in which we tested four size of samples ( 25, 50 , 75, 100) to illustrate the effect of changes in samples volume on parameters estimation , and for the sake of making a comparison between the used methods in estimation depend on the mean square error , and according to this results the maximum likelihood method is found to be the best and efficient way in estimation in which it gave the less mean square error, in addition to the models parameters by using this method was very close from the initial value that have been assumed to theparameters while the Bayesian method comes secondly in estimation Also this thesis included practical application dealing with the phenomena of earthquakes in Kirkuk province of which the time average was estimated by using maximum likelihood method and the Bayesian.

مقارنة طرائق تقدير معلمات ودالة معولية توزيع كاما ذي المعلمتين في حالة البيانات المفقودة باستخدام المحاكاة == Comparing The Estimation Methods of The Parameters And The Reliability Function For The Two Parameters Gamma Distribution In Case of Missing Data By Using Simulation

Author name: اوات سردار وادي
Supervisor name: ظافر حسين رشيد النجار
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: ان اغلب طرائق التقدير الاحصائية تفترض توفر بيانات تامة المشاهدات للعينات المدروسة، وقد بنيت جميع الطرائق على هذا الاساس. ولكن في الكثير من الظواهر الطبيعية، والاقتصادية، والاجتماعية وغيرها تتعرض جزء من بيانات هذه الظواهر الى الفقدان وتختلف اسباب الفقدان ف | Most of the statistical estimation methods depend on the availability of the complete data of the observations of the samples under study. All the statistical methods are based on this basis. However, part of the data of most of the natural, economic and social phenomena is prone to be missed for several reasons. The missing of the data may happen intentionally because of the high costs, risks or the lack of capabilities or unintentionally because of the failure of the recorders, the lack of the necessary requirements of production, the natural disasters, wars etc. Regardless of the various reasons, the incomplete data gives arise to a complex problem, which must be resolved by using statistical methods that deal with the incomplete data.In most cases, the failure times data of the individual component in the system has missing observations. Most of the reasons of having missing data go back to the meter, which registers the failure times of the whole system instead of a single component. Furthermore, the maintenance employees and the operators, who register the data, are responsible for maintaining the systems or the engines which fail to operate; they are not responsible for registering the data. Hence, it is not possible to have a convenient distribution of failure times because of the missing data of the individual component during the registration and because the available data represent the whole number of the failure times and the accumulative number of the operating. Consequently, the familiar methods of estimation are inconvenient. Therefore, some researchers derive and develop certain methods to estimate the parameters and Reliability Function using this kind of non - standard data for the various distributions of failure times.The research studies the Two Parameters Gamma Distribution, which is considered one of the most important, applicable and widely used distributions in the reliability realm and Survival Theory. It is mostly used as a model to distribute the failure times of the electrical, mechanical and electromechanical systems. The estimation of the parameters and the Reliability Function of this distribution in case of missing data has been made by using two important methods : the Maximum Likelihood Method and the Shrinkage Method. The former one consists of three methods to solve the MLE non - linear equation by which the estimators of the maximum likelihood can be obtained : Newton - Raphson, Thom and Sinha methods. Thom and Sinha methods are developed by the researcher to be suitable in case of missing data. Furthermore, the Bowman, Shenton and Lam Method, which depends on the Three Parameters Gamma Distribution to get the maximum likelihood estimators, has been developed. A comparison has been made between the methods in the experimental aspect to find the best method through simulation by using the Monte Carlo Method. Several experimentations have been made by using two of the important statistical measures : Mean Square Error (MSE) and Mean Absolute Proportional Error (MAPE). Generally, the developed Thom Method is found to be the best one for the Reliability Function estimation because it has the minimum Integral Mean Square Error (IMSE) and the minimum Integral Mean Absolute Percentage Error (IMAPE) in comparison with the other methods.

بناء توزيع احتمالي لوقت الفشل المستغرق باستعمال تحويل انتروبي لتوزيع Burr Type - XII == Building Probabilistic Distribution of The Failure Time Spent Using Entropy Transformation

Author name: اسيل نوري صالح
Supervisor name: اسماء غالب جابر الراوي
General topic: Administration and Economics
Specific topic: Statistics
Degree: Master
University: University of Baghdad - Faculty Of Administration And Economics - Department Of Statistics
Language: Arabic
University location: Baghdad
First pages:
Abstract: The research aims to find a probability distribution for the times of failure through function Entropy transformation using function reliability and cumulative distribution function, and after test data and found it follows the distribution of Burr TypeXII and experiencing volatility, were derived formula probability distribution of the new transfer application Entropy the probability distribution continuous Burr Type - XII and tested a new function and have found it to function probability Were derived SMA have found function probabilistic aggregate in order to adopt to generate data for the purpose of implementation of simulation experiments and research aims to compare estimate parameters probability distribution that has been extracted from the distribution formula using the methods of estimating possible Azam and the way White and estimation method proposed (appreciation mixed) and compare their adoption Metrology statistical mean square error and mean absolute error , has been using simulation for comparison between preference estimators and sizes different samples was conducted practical application to data from the General Company for electrical Industries was selected sample size (67) include electric motors until get failure in this setups.
1 ... 35 36 37 38 39 ... 53