MPhil Thesis
Browse
18 results
Search Results
Item The Role of High Leverage Points in Regression Diagnostics(University of Rajshahi, 2003) Khan, Md. Ashraful Islam; Imon, A. H. M. RahmatullahIn fitting a linear regression model by the least squares’ method, leverage values play a very important role. They often fo1m the basis of regression diagnostics as measures of influential observations in the explanatory variables. Much work has been done on the detection of high leverage values and a good number of diagnostic measures are now available in the literature. But neither of these methods is effective in the identification of high leverage points when multiple high leverage points are present in the data. In our study we proposed a new method for the identification of multiple high leverage points. The usefulness of this newly proposed method is studied under a variety of leverage structures through Monte Carlo simulation experiments. We also investigated the performance of the newly proposed method as a remedy to multi collinearity problem caused by the presence of multiple high leverage points.Item Comparison of Performances of Machine Learning Techniques in Healthcare Data(University of Rajshahi, Rajshahi, 2021) Maniruzzaman, Md.; Rahman, Md. JahanurDue to the increasing prevalence of diabetes and cancer, it is an urgent need to develop automated system that helps to detect disease using one of the modern technologies. Nowadays, Machine Learning (ML)-based methods have become very popular as an automatically model building techniques. Despite of the rapid development of theories for computational intelligence, application of ML-based classifiers to diabetes and cancer diagnosis remains a challenging issue. Still these ML-based classifiers did not give a satisfactory accuracy and therfore cannot correctly classify healthcare data like diabetes and cancer patients. Because most of the diabetes and cancer dataset are complex in nature and contains missing values, unusual observations, multi-collinearity problems and so on. In most of the existing research, the researcher did not use feature selection (FS) techniques to identify the risk factors of cancer and diabetes disease. They applied limited classifiers to classify and predict the diabetes and cancer status but they did not tune the hyper parameter of the classifiers, as a result, their accuracy and AUC were low. Thus, an attempt has been made in this study to increase the accuracy of the classifiers in diabetes and cancer data by considering the above factors in ML-based algorithm. The main objective of this study is to comparison the performances of ML-based methods in healthcare data and suggests the best model with better performance compared to the models published in the existing research.-----Item Robust Diagnostic Deletion Techniques in Linear and Logistic Regression(University of Rajshahi, 2008) Nurunnabi, Abdul Awal Md.; Nasser, MohammedIdentification of unexpected observations is a topic of great attention in modem regression analysis. At the beginning statisticians differ but now they recognize robust regression and regression diagnostics are two complementary remedies to study unusual observations. We use both of them for identifying irregular observations at a time. We find out the group deletion diagnostic methods that show better performance for identifying influential observations in linear regression. These are based on robust regression and/or relevant diagnostic methods so that these are free from huge computational tasks and reliable in presence of masking and/or swamping because of prior suspect-group identification. We find a technique that performs well in case of large number and high-dimensional data sets. We have done a classification task of unusual observations in linear regression according to their nature of consequences on the analysis, and model building process. At the same time the method performs well for identifying influential observations. This method may be a good addition to the existing graphical literature. We have seen that our proposed procedures in linear regression are also effective to the logistic regression after some modification and development to the existing identification techniques in linear regression. Our further contribution is to propose two new identification techniques for influential observations in logistic regression. The new methods show efficient performance for the proper identification of unusual observations and thereby provide less misclassification error in the response variable for the binomial logistic regression. Summarizing all the above issues we can say that we have made contribution in three areas: identification of influential observations in linear regression, classification of unusual observations in linear regression, and identification of unusual observations in logistic regression.Item A Study of Some Non-linear Stochastic Models on Renewable Resources Management: Application to Forestry(University of Rajshahi, 2010) Hossain, Md. Mowazzem; Shah, Md. AsaduzzamanManagement is a vibrant and multi phase dynamic process, just like social, scientific and engineering processes, that involves a wide spectrum of activities such as planning, coordination, communication, policy framing, decision making and their implementation. During the last three decades, the management of natural resources in general and that of renewable resources, in particular, has invited the attention of a large segment of researchers in various field [1-SJ. In order to maintain the ecological balance, as well as, meet the economic needs, the forest can play a vital role. Our government also taking special initiative to can-y forward a forestation program throughout the country. Also, to pollution free environment a forestation has no alternative. In this direction government organizations came forward to utilize the renewable resources for economic prosperity of the nation.Item Statistical Analysis of the Average Life of Electric Bulbs: A Comparative Study(University of Rajshahi, 1991) Islam, Md. Aminul; Mian, Md. Abul BasherThis thesis -- 'Statistical Analysis of the Average Life of Electric Bulbs: A comparative study' consists of six chapters of which chapter four and five are contributory, chapter six on the summary conclusion of the thesis. First three chapters of the thesis are introduction and discussions on the selection of a life testing model and the underlying methods of statistical inferences which has been used in the subsequent chapters towards the contribution in the' thesis. The introductory chapter one contains a statement of the problem we have undertaken a brief review of earlier studies and exploration of the possibilities of further work in relation to the present study, aims and scope of the study. It also provides a brief discussion on the concept of life testing and reliability and distributions of life times. A survey of some basic life testing models which has been used in contributory chapters has also been appended in this chapter. Chapter two of this thesis is devoted on discussions on the selection of a. life testing model to suit the analysis and prediction for a particular set of data. Almost all life testing data available for analysis and prediction are incomplete or censored. Most of the tools and techniques available in life testing literature for discriminating between competing life testing models are sensitive to the nature and size of censoring……………………..Item Anthropometric Study of the Shantal Community in Rajshahi District(University of Rajshahi, 2006) Karim, Md. Rezaul; Islam, Md. Nurul; Ali, Md. AyubThe aim of the thesis was to study Anthropometric variables staure, body weight, body mass index (BMI), sitting height and chest circumference, to find their relationships with some of the demographic and Socio-economic variables of Shantal community in Rajshshi District, Bangladesh. The_demographic and socioeconomic variables were collected through a questionnaire (in Appendix-I). Anthropometric measurements were taken by the· author himself. The cluster sampling technique was applied. The sample size were 396 for male and 438 for female. The present study demonstrated that the average stature, body weight, sitting height and chest circumference of young Shantals were increasing comparatively with those of oldest Shantals. Almost every Shantals were very lean and thin and stature, body weight, sitting height, and chest circumference of Shantals were positively and significantly related to each other. The present study indicated that the female onset of menarche of Shantals community reached earlier who were heavier and became late menarche who were both taller and thinner. Also lower BMI of female Shantals were reached age at menarche earlier than those of larger BMI. Also, age at menarche was earlier for those whose family income were higher and was late for those whose family income were lower and the onset of menarche reached earlier than those who have fewer number of sibling. Through the logistic regression, the present study suggest that the stature of literate male Shantals were shorter than the stature of illiterate Shantals and their differences were signHicant (p<0.05). The body weight and occupation were negatively associated but their differences were insignificant. The stature of literate female Shantals were higher than the stature of illiterate female Shantals and the body weight of female labors were lower than the body weight of non-labor female Shantals with significant differences. Sexual differences among stature, body weight and BMI of Shantals were found and the male Shantals were taller in stature and heavier in body weight than female Shantals and female Shantals were lean and thin. Government should take favorable view and necessary co-operations for their better health, social and economic management.Item Factors Affecting Infant and Child Mortality in Bangladesh: A Multivariate Analysis(University of Rajshahi, 2008) Islam, Md. Mogibul; Islam, Md. NurulThe reduction of infant and child mo11ality in the developing countries is one or the most substantial achievements of human kind. However, in spite of various effective intervention programs, the infant and child mortality are considerably high in Bangladesh. Therefore, in this study, an attempt has been made to assess the levels, patterns and determinants of mortality in Bangladesh utilizing nationally representative data from Bangladesh Demographic and Health Survey (BDHS) 2004. The purpose of this study is also to identify user-related factors, which influence the infant and child mortality in Bangladesh. Differential pattern in infant and child mortality in Bangladesh is examined using bi variate analysis, logistic regression analysis and also factor analysis in multivariate approach. The study results show that several socioeconomic, demographic and household variables affect infant and child mortality. These are: place of residence, division, mother's education, father's education, father's occupation, age of mother at birth of child, sex of child, preceding birth interval, availability of electricity, sources or drinking water, toilet facilities, exposure to mass media i.e., radio, TV and floor/wall/roof materials. Multivariate analysis results indicate that type of place of residence, sex of child, mother's education, father's occupation and division are important factors that have significant influence on infant and child mortality. Ti1e most significant predictors of neonatal, post-neonatal and infant mortality are mother's education and father's occupation. Father and mother's education both are persistent socioeconomic predictors of mortality. Construction materials of wall of houses are found significant for neonatal mortality. Floor materials are found significant for infant mortality. Furthermore, differential analysis shows that, male children experienced substantially higher mortali1y than female children did at neonatal, infantile and child periods but in post-neonatal period the relationship is opposite.Item Investigating the Trend and Impact of Some Variables on HYV Boro Rice Production in Northern Region of Bangladesh(University of Rajshahi, 2011) Kundu, Ranjan Kumar; Karmokar, Provash KumarBoro rice is playing a significant role in the total rice production of Bangladesh. High yielding variety (HYV) Boro rice is grown enormously in the Northern region of Bangladesh. On the basis of per hector production, three HYV Boro rice producing districts have been selected by Simple Random Sampling for the present study. The selected three districts are Rajshahi, Rangpur and Dinajpur. To fulfill the objectives of the present study Trend Analysis and Factor Analysis have been anticipated. Suitability of commonly used trend models like Linear, Compound and Quadratics trend models for the selected districts were checked by Mann-Kendall and Spearman non parametric statistical tests and parameters of these models were estimated by OLS technique. The DW statistics are obtained by OLS technique shows the evidence of autocorrelation in the data sets. To overcome this problem we have used CochraneOrcutt technique to estimate of parameter of the trend models. The Cochrane-Orcutt technique minimizes the autocorrelation problem. The JB test shows the normality assumption of errors for all the models. Although the R2 and adjusted R2 have increased for all the models it is remarkable of about 17%-18% increasing of these values for Rajshahi districts in both the Linear and Compound models. The HYV Boro productions of the selected districts have been forecasted up to year 2021 for three trend models using the data from 1971 to 2003 by Cochrane-Orcutt technique and the results are given in detailed in Chapter Three. It is seen from the result that the Compound Trend model performs better maintaining the growth rate 1.009 for Rajshahi, 1.012 for Rangpur, 1.008 for Dinapur and its forecasting performance is the best among the trend models. Production of HYV Boro rice may depend on several climatic and non climatic variables like, rainfall, temperature, wind speed, humidity, cloud coverage, CO2 emission, fertilizer use and tractor use. In this study, we are also interested to identify the factors whose impacts are influential to the HYV Boro production of the selected districts of northern region of Bangladesh. Factor analysis, a data reduction technique have been used to identify the impact such variables on HYV Boro rice production in northern region of Bangladesh. Our analysis has identified five influential factors for the HYV Boro rice production for each of the districts. Therefore it is shown that the selected factors account for 82.33% of the total variance for Rajshahi district, 75. 71 % of the total variance for Rangpur district and 71.57% of the total variance for Dinajpur districts. The findings of this study convey a message to the agriculturist, concerned authorities and policy makers that keep watching of climatic factors and proper handling of other relevant factors in Boro rice production of northern region that would be significant in food supplying in this region and enhance the food security of Bangladesh.Item Analysing the Variability of Rainfall Threshold Values for Drought Prognosis in Bangladesh(University of Rajshahi, 2013) Alam, Md. Mahbub; Rahman, M. Sayedur; Hossain, Md. RipterThe Rainfall plays a significant role in the agriculture of country. Rainfall is the most important weather parameter affecting non-irrigate crop areas for Bogra, Jessore and Feni water deficits and excess water are the greatest constraints for rainfall rice yields in this region. The daily rainfall data for 30 years during the period 1981-2010 of the rainfall stations in Bogra, Jessore and Feni are considered in this study. The daily data were reduced in the weekly form and the drought index has been calculated using the probability of Markov chain model. Drought is temporary but complex feature of the climate system. Agricultural drought is mainly concerned with inadequacy of rainfall. Markov chain model have been used to evaluate probabilities of getting a sequence of wet-dry weeks over this region. An index based on the parameters of this model has been suggested for agricultural drought measurement in this region. The results of our Bogra district annually rainfall data we observed the moderate drought, pre-kharif rainfall data we observed the mild occasional and moderate drought, kharif rainfall data we observed the occasional and rabi rainfall data we observed the chronic prone. As a result failure of rains and the occurrence of drought during any particular growing season lead to severe food shortages. A Markov Chain model is established to fit daily rainfall data for the various aspects of rainfall occurrence patterns and could be mathematically derived from the Markov Chain by using maximum likelihood estimate and these were also established to fit the observed data. The rainfall probability was not found to very much during rabi (November- February) season. But much more variation of rainfall probabilities was observed during both kharif (June-October) and pre-kharif (March- May) seasons. Obviously, one would presume conditions variation in these probabilities also yearly, seasonally and annually. Based on these findings and using chi-square test, it could be concluded that the model fit was good. This study investigates the methods to obtain estimates of the conditional probabilities, the probability of success and its probability distribution to describe the yearly, seasonal and annual variability. The limited data set the results are quite good and the model is doing a reasonably good job of daily rainfall.Item Study of Potential Impacts of Climate Change on Food Production in Bangladesh(University of Rajshahi, 2013) Chowdhury, Md. Mizanur Rahman; Rahman, M.Sayedur; Hossain, M. RipterThe climate plays significant role in the agriculture of a country. The Rajshahi region is the driest part of the country in terms of rainfall. Rainfall is the most important weather parameter affecting non-irrigated crop areas. Water deficits and excess water are the greatest constraints for rain fed rice yields in this region. The daily rainfall data for 30 years during the period 1981-2010 of the station Rajshahi, Barisal and Comilla district in region are considered in this study. This study also considered weekly rainfall for 2.5mm and 5mm. The daily data were reduced in the weekly form and the drought index has been calculated using the probability of Markov chain model. Drought is temporary but complex feature of the climate system. Agriculture drought is mainly concerned with inadequacy of rainfall. Markov chain model have been used to evaluate probabilities of getting a sequence of wet-dry weeks over this region. An index best on the parameters of this model has been suggested for agriculture drought measurement in this region. The results indicate that the drought index for rainfall 2.5mm of rain for Rajshahi district annual is Mild, pre-kharif are Mild, Occasional, and Moderate, kharif is Occasional and Rabi is Chronic. The drought index for rainfall 5mm of rain for Rajshahi district annual is Mild, pre-kharif are Mild, Occasional and Moderate, kharif is Occasional and Rabi is Chronic. The drought index for rainfall 2.5mm of rain for Barisal district annual is Mild, pre-kharif are Mild, Occasional and Mild, kharif is Occasional and Rabi is Chronic. The drought index for rainfall 5mm of rain for Barisal district annual is Mild, pre-kharif is Occasional, kharif is Occasional and Rabi is Chronic. The drought index for rainfall 2.5mm of rain for Comilla district annual is Mild, pre-kharif are Mild, Occasional, Mild and Moderate, kharif is Occasional and Rabi is Chronic. The drought index for rainfall 5 mm of rain for Comilla district annual is Occasional, and Mild, pre-kharif are Mild, Occasional and Mild, kharif is Occasional and Rabi is Chronic. As a result, failure to rains and the occurrences of drought during any particular growing season lead to severe food shortages. A Markov chain model is established to fit daily rainfall data for the various aspects of rainfall occurrence patterns and could be mathematically derive from the Markov Chain by maximum likelihood estimate and these were also established to fit the observed data .The distribution of the number of success is asymptotically normal. The rainfall probability was not found to very much during rabi (November- February) season. But much more variation of rainfall probabilities was observed during both kharif (June-October) and pre-kharif (March- May) seasons. Obviously, one would presume conditions variation in these probabilities also yearly, seasonally and annually. Based on these findings and using chi-square test, it could be concluded that the model fit was good. This study investigates the methods to obtain estimates of the conditional probabilities, the probability of success and its probability distribution to describe the yearly, seasonal and annual variability. The limited data set the results are quite good and the model is doing a reasonably good job of daily rainfall. The probability can be used for both instantaneous and climate time scale retrievals. As the distribution of number of success asymptotically normal, it is playing a vital role for important decisions such as disaster prevention preparedness strategy. This study will contribute toward a better understanding of the climatology of drought in major monsoon region of the world. The findings of this study will be helpful for every researcher.