Browse
Recent Submissions
Item Predicting Returns on Equity using various Shrinking methods(UMT.Lahore, 2019) Shabnam AmreenHigh prediction accuracy and relevant variable selection are two vital objectives in statistical learning. Various shrinkages methods such as Lasso, Adaptive Lasso, Elastic-net and Generalized Elastic-net have been suggested in the literature. Unlike discrete methods, e.g. subset selection, the continuous shrinkage while achieving optimal prediction and doing consistent variable selection simultaneously is the desirable properties of these shrinkage methods.Item Exploring Maternal Risk Factors for Congenital Hydrocephalus(UMT.Lahore, 2019) Munawar GhousIn this study we investigate different factors that were associated with congenital hydrocephalus. Data has been collected from The Children’s Hospital & Institute of Child Health Lahore and from Lahore General Hospital Lahore. Sample size was 180 (120 control and 60 cases). Binary Logistic Regression analysis has been used to explore the factors that were associated with congenital hydrocephalus. It is concluded that residential area of mother, educational level of mother, family history of hydrocephalus, family history of tuberculosis (TB), cousin marriage, smoking habit of father, easy approach to gynecologist, use of folic acid during pregnancy, diabetes to mother during pregnancy, hypertension to mother during pregnancy, maternal infection, neurological problem to mother, consistent headache to mother during pregnancy, nausea to mother during pregnancy and social economic status were the significant risk factors that were associated with progression of congenital hydrocephalus in newborn infant.Item Generalized regression-exponential type ratio-product estimators(UMT.Lahore, 2019) ASMA MASOOD MALIKIn this thesis, Generalized regression-exponential type ratio-product estimators under Simple Random and Ranked Set Sampling have been proposed for estimating finite population mean, using the information form auxiliary variable, under simple random sampling, ranked set sampling and median ranked set sampling. In chapter 1, the simple random sampling, ranked set sampling and median ranked set sampling designs have been discussed. The types of ranking and modification of ranked set sampling with respect to exponential, ratio, regression and product estimators have been given. The previous work related to simple random sampling, ranked Set Sampling and median ranked set sampling has been given in proceeding chapter along with the use of auxiliary information. Whereas Chapter 3 contains the existing estimators for the population mean which has been developed in simple random sampling, ranked set sampling and median ranked set sampling. In Chapter 4 the bias and mean square error expressions have been derived for the estimator under in simple random sampling, ranked set sampling and median ranked set sampling. In chapter 5 empirical studies were conducted. In which Four populations were taken from “vincentarelbundock”. i.e. Crohn's disease data, performance of computer CPU, Auto dataset, and Morphological Measurements on Leptograpsus Crabs. From which we ensured proposed estimator under simple random sampling, ranked set sampling and median ranked sampling are highly efficient than existing estimators (Cochran (1977), Mohanty (1967), Samawi (1996), Muttlak (2000), Hanif and samiuddin (2006) chain ratio estimator 1 and 2, Hanif and samiuddin (2006) regression-ratio-product estimator I,II,III, Hanif et al (2009), and Gajendra (2017)). Especially estimator under median ranked set sampling.Item Exploring the Determinants of Education, Wealth Index and Dwelling Status of Household Heads in Punjab Multiple Indicators Cluster Survey (MICS)(UMT.Lahore, 2019) AZKA SHAMSHADIn our study, we explore various factors influencing the Education, Wealth Index and Dwelling Status of Household (HH) Heads in Punjab Multiple Indicators Cluster Survey (MICS) 2017-18. Two Ordinal Logistic Regression models and one Multinomial Logistic regression model are proposed to find the significant effect of independent variables on each of the response variables individually. It is concluded that as the education level of HH that belongs to rural area is significantly higher than the head of the household that belongs to the urban area. The education level of male HH head is significantly higher than the head of the female HH head. The education level of HH head who speaks Urdu is significantly higher than that of HH head who speaks other language; the education level of HH head who speaks Saraiki is also significantly higher than that of HH head who speaks other language. The education level of HH head that avail Inter-connected electricity facilities are significant than those of lower level of education of HH head and with off grid electricity facilities times more significant than those of lower level of education of HH head. The education levels of HH head that having internet facilities are significantly higher than those of without internet facilities. The education level of HH head that are having their own house are less significant than that of HH head who have other dwelling status and the education level of HH head that are having rented house are also less significant than that of HH head who have other dwelling status. The education level of HH head that belongs to the poor status is significantly lower than that of HH head that belongs to the richest category, the education level of HH head that belongs to the secondary level is significantly lower than that of HH head that belongs to the richest category, the education level of HH head that belongs to the middle level is significantly lower than that of HH head that belongs to the richest category and the education level of HH head having fourth level is significantly lower than that of HH head that belongs to the richest category. It is concluded that for 2nd model, the response variable Wealth Index Quintile conclude that number of HH members, Education level of HH, Area of HH, Sex of HH, Electricity facility, Internet access at home and the dwelling status of HH are significantly higher than their reference categories. In the variable Language of HH head, all categories except English Language of HH are significant. In Multinomial Logistic Regression model, the response variable has been divided into three categories. First half of the table describe the results of the Own category of Dwelling status while the 2nd half of the table describe the results of the Rent category of Dwelling status and the 3rd category other is taken as reference category. First half of the table displays that all of the independent variable i.e. number of HH members, Area of HH, Sex of HH, HH have electricity, Wealth Index Quintile showing significant results while the preschool and secondary level of Education of HH head are significant and in the variable Language of HH, only Saraiki is showing the significant result. Whereas 2nd portion shows that all the predictors’ i.e. number of HH members, Area of HH, Sex of HH, HH have electricity, Wealth Index Quintile showing significant results while the Language of HH, the category English is not showing the significant results. In Electricity, the category off-grid is showing significant result. Internet access at home is also not significant. Estimating response probabilities are also calculated to check the occurrence of dependent variable in specific category.Item Estimators Under Stratified Ranked Set Sampling Using Auxiliary Information(UMT.Lahore, 2019-02-22) Rahila AkhtarIn this thesis, exponential and ratio estimators have been proposed for estimating finite population mean, using the information form auxiliary variable, under stratified ranked set sampling design. In chapter 1, the simple random sampling, ranked set sampling and stratified ranked set sampling designs have been discussed. The types of ranking and modification of ranked set sampling with respect to exponential and ratio estimators have been given. The previous work relating to simple random sampling, ranked Set Sampling and stratified ranked set sampling has been given in chapter 2 along with the use of auxiliary information. Whereas Chapter 3 contains the estimators for the population mean which have been developed in ranked set sampling and simple random sampling. The proposed estimators are discussed in Chapter 4. The bias and mean square error expressions have been derived for the estimators under stratified ranked Set sampling. Simulation study based on Monte Carlo technique is conducted in chapter 5. The efficiency comparison of the proposed estimators with the competitor estimators are performed. It has been founded that the proposed estimators are more efficient than competitive estimators under simple random sampling and ranked set sampling design.Item Mixture Regression Ratio Product Estimator under Stratified Random Sampling(UMT.Lahore, 2019-03-28) Iqra JavedIn this study, Mixture Regression Ratio Product Estimators for single phase sampling under stratified random sampling have been proposed, by incorporating the simultaneous use of information on auxiliary variables and attributes. The estimators have been proposed for two different cases and their mean square errors have been derived mathematically. A Simulation study has been done by using simulated data, to check the distribution of proposed estimators. This study shows that proposed estimators, seems to follow the normal distribution. An empirical study has also been done by considering two natural data sets. Mean square errors (MSE’s) for the proposed estimators have also been computed and Efficiency comparisons made with single phase mixture regression estimators proposed by Moeen et al. (2012). On the basis of MSE’s computed through simulation and empirical studies, it is to be concluded that proposed estimators are more efficient than that of estimators proposed by Moeen et al. (2012) for simple random sampling.Item Survival Analysis on Chronic Kidney Disease(UMT.Lahore, 2019) MAHRUKH SHAZADIIn this research work the sample of the 94 patients including 43 males 51 and female patients have been studied to explore the effects of diabetes and high blood patients on survival time. Kaplan Meier non-parametric technique is used to determine survival. In Kaplan Meier technique survival curve is used to determine a fraction of patients surviving a specified event, like death during given period of time. We are also estimated median survival time for male and female and construct 95% confidence interval. Cox-regression technique is also applied to observe the effect of significant factors on survival time. These factors include gender, age, protein in urine, glomerulonephritis, hospitalized, stages of CKD, status on exit date, time Diagon, kidney stone, diabetes, high blood pressure, due to hospitalization, Diga ultrasound out of these factors only kidney stone, hospitalize, stages of chronic kidney disease, diabetes. It is also observing from the study that the survival time of the patients is reduce those are suffering from chronic kidney disease.Item ESTIMATION OF THE POPULATION MEAN UNDER RANKED SET SAMPLING(UMT.Lahore, 2019-02-22) HUMAIRA LATIFIn this thesis, exponential ratio and simple type estimators have been suggested for estimating finite population mean, using the information from auxiliary variable, under Ranked set sampling procedure. In chapter 1, the explanation about the ranked set sampling design and its modification with respect to exponential and simple ratio estimator has been given. The effect of ranking errors during ranked set sampling has also been discussed in this chapter. The previous work related to Ranked Set Sampling has been given in proceeding chapter 2 along with the use of auxiliary information in ranked set sampling design. In chapter 3 some existing estimator have been re-produced along with Mean Square Errors. Following the above chapters, the major contribution of this dissertation that leads to the epitome of the entire topic appears from the Chapter 4, three estimators are Proposed to estimate the population mean using different central tendencies of single auxiliary variable. Their Mean Square Error, Percentage Relative Efficiency and Biases are discussed in this chapter. Simulation study based on Monte Carlo technique is conducted in chapter 5. The efficiency comparison of the proposed estimators with the competitor estimators are performed. It has been founded that the proposed estimators are more efficient than competitive estimators under simple random sampling and ranked set sampling design. In chapter 6 theoretical results are supported with the help of three real life populations.