Browsing by Author "Jolayemi, E.T."

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    A Comparative Study on Zero-truncated Generalized Poisson-Lindley and Zero-truncated Poisson-Lindley Distributions
    (International journal of Mathematical Archive (IJMA), 2017) Aderoju, S.A.; Jolayemi, E.T.; Ibrahim, A.O.
    In this paper, Zero-truncated Com-binomial distribution was derived and investigated its behavior in modeling structurally non-zero data. The proposed distribution is characterized by two parameters, which make it flexible. The maximum likelihood method is used to obtain the estimators of the parameters through R-software. Two real-life datasets were used to evaluate its performance. The statistic (chi square goodness-of-fit) with the p-value shows that the proposed Zero-truncated Com-binomial distribution yields “a good fit”.
  • Thumbnail Image
    Item
    Issues of Class Imbalance in Classification of Binary Data: A Review
    (International Journal of Data Science and Analysis, 2019) Aderoju, S.A.; Jolayemi, E.T.
    Handling classification issues of class imbalance data has gained attentions of researchers in the last few years. Class imbalance problem evolves when one of two classes has more sample than the other class. The class with more sample is called major class while the other one is referred to as minor class. The most classification or predicting models are more focusing on classifying or predicting the major class correctly, ignoring the minor class. In this paper, various data pre-processing approaches to improve accuracy of the models were reviewed with application to terminated pregnancy data. The data were extracted from the 2013 Nigeria Demographic and Health Survey (NDHS). The response variable is “terminated pregnancy” (asking women of reproductive age whether they have ever experienced terminated pregnancy or not), which has two possible classes (“YES” or “NO”) that exhibited class imbalanced. The major class (“NO”) is 86.82% (of the sample) representing Nigerian women of age 15 – 49 years who had never experience terminated pregnancy while the other category (minor class) is 13.18%. Hence, different resampling techniques were exploited to handle the problem and to improve the model performance. Synthetic Minority Oversampling Technique (SMOTE) improved the model best among the resampling techniques considered. The following socio-demographic factors: age, age at first birth, residential area, region, education level of women were significantly associated with having terminated pregnancy in Nigeria.
  • Thumbnail Image
    Item
    Modelling Of Accident Data In Nigeria: Poisson And Poisson Gamma Regressio
    (2015) Aderoju, S.A.; Jolayemi, E.T.
    Poisson and Poisson-gamma (Negative Binomial) regression models belong to the group of generalized linear models that are suitable to model count data. While the two regression types often give similar results, there can be differences in the effects of the covariates they estimated depending on weather the data is equi-dispersed, over-dispersed or under-dispersed. This work therefore demonstrates the behaviours of Poisson and Poisson-gamma regression models when the data is overdispersed (variance greater than the mean). The two regression types were employed to model the number of people killed in road accidents as a function of the number of fatal, serious and minor accidents that occurred. Results from the two regression models indicated that number of people killed in road accidents is determined majorly by the number of fatal and serious cases of road accidents that occurred and not really by the number of minor accidents. The differences in the two regression models are described in the light of the over-dispersion in the data and the ability of the models to account for it. Road accident data of Kwara State, Nigeria for nine consecutive years from 2000 to 2008, primarily collected by the Nigeria Police Force (NPF), were employed in this study.
  • Thumbnail Image
    Item
    On modeling Zero-truncated count data
    (Journal of Global Research in Mathematical Archives, 2018) Aderoju, S.A.; Jolayemi, E.T.
    In this paper, a more generalized zero-truncated distribution that come from the mixture of distributions was developed and implemented for non-zero count data. The application of the distribution was demonstrated and its performance assessed against some existing ones using real life datasets. Following the idea of mixed distributions, the Zero-Truncated Com-Binomial (ZTCB) is from the mixture of Conway-Maxwell-Poisson type generalization to the Binomial distribution. The first two moments via probability generating function were also derived. The Maximum Likelihood Estimations of the parameters were also obtained by direct maximization of the log-likelihood function using “optim” routine in R software. The findings of the study showed that: the ZTCB distribution is more robust to handle all levels of dispersion than Zero-Truncated Multiplicative Binomial distribution. The statistics (chi square goodness-of-fit) as well as Deviance (in example four) and the AIC show that the proposed ZTCB distribution yields best result among the models under consideration. This paper therefore provides useful alternative to the existing zero-truncated distributions.
  • Thumbnail Image
    Item
    On Zero-Truncated Negative Binomial with Excess Ones
    (Asian Journal of Probability and Statistics, 2023-05-11) Jolayemi, E.T.; Aderoju, S. A
    In this paper, Zero-truncated negative binomial distribution is modified to include excess ones to improve goodness-of-fit. This is necessary when data are dispersed and zero has been eliminated from data structurally. However, when the ones are unduly large, the proportion of this excess must be recognized and estimated to improve the fit. This development is applied using real data from a national survey.
  • Thumbnail Image
    Item
    Power Hamza distribution and its applications to model survival time
    (Journal of the Nigerian Statistical Association, 2022) Aderoju, S.A.; Jolayemi, E.T.
    In this paper, a new lifetime model for survival time analysis has been introduced, which is called power Hamza distribution (PHD) that generalizes the Hamza distribution. Some properties of the new distribution, including the probability density function, cumulative distribution function, moments, failure rate function or hazard function and moments were presented. The model provides more flexibility than the Lindley and Hamza distributions in terms of the shape of the density and hazard rate functions. Estimate of the model parameters were obtained via the method of maximum likelihood and applications of the model were made to two real data sets. By using some criteria like Akaike Information criteria (AIC) and Bayesian Information Criteria (BIC) and other statistic, the PHD model provides better fits than other classical lifetime models such as Hamza, exponential and Lindley distributions.