Browsing by Author "Omekam, Ifeyinwa"

Now showing 1 - 10 of 10
  • Item
    Bayesian Additive Regression trees for Predicting Colon Cancer: Methodological Study (Validity Study)
    (Turkiye Klinikleri (10.5336), 2022-06-26) Olaniran, Oyebayo; Olaniran, Saidat; Popoola, Jumoke; Omekam, Ifeyinwa
    Objective: The occurrence of colon cancer starts in the inner wall of the large intestine. The survival of colon cancer patients strongly relies on early detection. Diagnosing colon cancer using clinical approaches often takes longer, especially in most developing countries with limited facilities. The recent use of microarray technology has presented a new approach for the oncologist to diagnose cancer cells using non-clinical machine learning methods. In this paper, the aim is to predict the status of colon cancer tissues using the Bayesian Additive Regression Trees (BART) and 2 other machine learning methods. Material and Methods: The development and comparative analysis of BART alongside 2 other competing methods (Random Forest: RF and Gradient Boosting Machine: GBM) were implemented. The dataset used for the analysis is the microarray colon cancer data which consists of 2,000 gene expression measurements for 62 tissue samples. Results: The methods are compared based on overall metrics (accuracy, balance accuracy, detection rate, F-measure and AUC) and class-specific metrics (sensitivity, specificity, positive predictive value and negative predictive value). The overall metrics results showed that the best method is RF. The class-specific metrics results showed that BART is better than RF. Conclusion: On average, BART is more sensitive in detecting the presence of colon cancer cells, while RF is more accurate and specific in detecting the presence or absence of colon cancer cells.
  • Item
    Cubic Transmuted Lindley as a Better Distribution in the Family of Rayleigh Distributions
    ("Tibiscus" University of Timişoara, România, 2020-06) Adefowope, Adesiyan; Omekam, Ifeyinwa; Adejumo, Adebowale
    A new statistical model for non-normal data is proposed with some of its statistical properties as Cubic of the Transmuted Lindley distribution (CTLD), based on a new family of life time distribution. The new model contains some of lifetime distributions as special cases such as exponentiated Lindley, transmuted Lindley and Lindley distributions. These include the density and hazard rate functions with their behavior, moments, and moment generating function, skewness and kurtosis measures. Maximum likelihood estimation of the parameters and their estimated are derived. An application of the model to a real data set is presented and compared with the fit attained by other well-known existing distributions.
  • Item
    Derivations for the Families of Generalized Distributions
    ("Tibiscus" University of Timişoara, România, 2017-12) Omekam, Ifeyinwa; Adejumo, Adebowale
    Generalization of distributions is usually motivated by limitations in characteristics of existing distributions so as to introduce more flexibility and improve goodness of fit. This is done by parameter induction into an existing distribution and therefore remains an approach to generalizing distributions. In this article, families of generalized distributions are generated by sequential application of methods in permutations of five distinct parameter induction methods: Lehmann Alternative 1 (LA1); Lehmann Alternative 2 (LA2); Marshal and Olkin Method (M-OM); α-Power Transformation (APT); and Power Transformation Method (PTM). This is done by taken two methods at a time. Sixteen distinct families of generalized distributions were generated. Some of the families of generalized distributions obtained are already in existence while several others are entirely new.
  • Item
    Factors Influencing the Choice of Place of Delivery of a First Child among Nigerian Women
    (College of Natural and Applied Sciences, University of Dar es Salaam, 2022-06-17) Adeniyi, Olakitan; Afolabi, Nathanael; Akinrefon, Adesupo; Omekam, Ifeyinwa; Olanijolu, Ifeoluwa
    Among the factors responsible for maternal deaths in developing countries is complications during pregnancy and childbirth, and as such, the choice of place for delivery is vital as the quality of attention received can either aid or reduce the risks attached to child bearing. This study seeks to statistically model factors that determine the place of delivery in Nigeria among women of reproductive age taking into consideration possible nesting structure in the mode of data collection. A two-level hierarchical multilevel logistic model with the individual women as the lower level and the state of residence as the second level was applied to 2018 Nigeria Demographic Health Survey (NDHS) data. Results showed that the odd of choosing a government or private facility reduced by 58.6% and 85.9%, respectively. Women in the rural areas are 12% and 20% less likely to choose either a government or private facility, additional education increases the odd of choosing any of the two facilities against home delivery. Other significant variables were wealth index, religion, assistance during labour and the number of antenatal visits.
  • Item
    Flexibility of Generalized Distributions
    (Royal Statistical Society Nigeria Local Group, 2022-09-12) Omekam, Ifeyinwa; Adejumo, Adebowale
    Some existing distributions are limited in shapes of Probability Density Function (PDF) and Hazard Function (HF) which constrains their use in analysis of certain types of data. Generalizing these distributions often deal with this constraints on usage by introducing flexibility. Generalized distributions were derived using the Generalized Pareto Distribution (GPD) as base distribution. Exponentiated GPDs called Lehmann Type II GPD (LIIGPD) and Lehmann Type I GPD (LIGPD) having an additional parameter each were obtained by applying Lehmann Alternative 1 (LA1) and Lehmann Alternative 2 (LA2) parameter induction methods respectively. Flexibility of generalized distributions was established by comparing the shapes of probability density and hazard functions of LIIGPD and LIGPD with those of the GPD. No new probability density or hazard shape was introduced by LIIGPD but the new shape introduced by LIGPD demonstrated flexibility of generalized distributions. Generalized distributions do not always introduce new density and hazard shapes but often improve flexibility of distributions.
  • Item
    Goodness of Fit Tests for Some Generalized Distributions
    (Federal University Wukari, Taraba State, Nigeria, 2020-04) Omekam, Ifeyinwa; Adejumo, Adebowale
    Analysis of data without a pre-knowledge of the distribution that describes data may lead to misleading or irrelevant result. Distribution fitting to data often lead to the selection of the best fitting distribution for data analysis. In this article, some existing distributions are fitted to Maternal Mortality Ratio (MMR) data. Using the best fitting distribution obtained as base distribution, generalized distributions having additional parameters are then derived and subsequently fitted to MMR to assess goodness of fit. Generalized distributions improved goodness of fit.
  • Item
    Goodness of Fit Tests for Some Generalized Distributions
    (Federal University Wukari, Taraba State, Nigeria., 2020-04) Omekam, Ifeyinwa
    Analysis of data without a pre-knowledge of the distribution that describes data may lead to misleading or irrelevant result. Distribution fitting to data often lead to the selection of the best fitting distribution for data analysis. In this article, some existing distributions are fitted to Maternal Mortality Ratio (MMR) data. Using the best fitting distribution obtained as base distribution, generalized distributions having additional parameters are then derived and subsequently fitted to MMR to assess goodness of fit. Generalized distributions improved goodness of fit.
  • Item
    Modified Frechet Distributions and Their Generalized Families
    (Faculty of Science, Kaduna State University, 2022-07-10) Omekam, Ifeyinwa; Adeniyi, Olakitan; Adejumo, Adebowale
    The Frechet distribution is used for modeling extreme events. There are different approaches to developing statistical distributions which include the use of translation methods, system of differential equations, quantile methods among others. Existing statistical distributions are also modified or generalized to accommodate other different types of data and improve goodness of fit to data. Addition of extra parameter(s) is one approach used for generalizing existing distributions such that the base distributions are embedded in the new generalized distributions. Some methods of parameter induction were used to obtained families of generalized distributions. Parameter(s) were also introduced into the probability distributions of the Frechet distribution to derive functions of its modified versions belonging to each of the generalized families derived. Further study is recommended on some of the modified Frechet distributions and their generalized families.
  • Item
    Regional Variation in Age at First Marriage among Women of Reproductive Age in Nigeria
    (Kwame Nkrumah University of Science and Technology, 2023-04-20) Adeniyi, Olakitan; Olonijolu, Ifeoluwa; Omekam, Ifeyinwa; Akinrefon, Adesupo; Oladuti, Olubimpe
    The society and the environment have great influence on the attitudes and the decisions made by her residents. The age at which a woman enters marriage to some extent is influenced by the society and the environment she lives in and some other socio-demographic factors. This study employs a hierarchical survival analysis which account for state differences in the age at first marriage among Nigerian women using the dataset from 2018 National Demographic Health survey. The Cox model with two independent random effects was used to provide parameter estimates as well as estimates of the random effects variances at all the levels. It was found that state heterogeneity had the highest contribution and location of residence within the state also contributed to the differences in the timing of marriage. The study also revealed that region, location of residence, wealth index, respondent age at first sex, birth cohort, religious affiliation and educational qualification of the women were significant factors in determining the age at first marriage.
  • Item
    Some Extended Pareto type I Distributions
    (Obafemi Awolowo University, Ife, Osun State, Nigeria under the platform of Africans Journals Online, 2022-10-13) Omekam, Ifeyinwa; Popoola, Jumoke; Gatta, Nusirat; Adejumo, Adebowale
    Probability distributions are essential in data modeling. Introduction of parameter(s) into existing probability distributions is a method of extending or generalizing distributions to produce more flexible distributions and for better fit to data. The Pareto type 1 distribution (PT1) is a right skewed continuous distribution originally used in description of wealth and income but also used for modeling other right skewed data. To add flexibility, Pareto type 1 distribution was extended by introducing parameter(s) into its probability distribution to accommodate more types of data. Some functions of the extended Pareto type 1 distributions were derived using five parameter induction methods. Flexibility of extended distributions was demonstrated through comparisons of density and hazard function shapes of some of the extended distributions with those of the PT1. Further study on properties of non-existing extended Pareto Type I distributions and real-life applications are recommended.