Comparison of different distributions to choose the best distribution to estimate the survival function for corona patients
A Thesis Submitted to
Council of The Administration and Economics/ Karbala University as Partial fulfillment of the Requirements for the Degree of Master of Science in Statistics
Presented by researcher
Iqbal Qasim Ramadhan
Supervised By
Ass. Prof. Raissan Abdulimam Zalan
Abstract
The study aims to find the appropriate probability distribution to represent and describe the survival data for a sample of patients infected with Coronavirus (Covid-19) from the date of their admission to the hospital until death measured in days, as well as estimating the survival function of the aforementioned sample.
The simulation experiment was conducted using the Monte Carlo method with different sample sizes (30, 50, 100, 200) for some basic probability distributions (Exponential distribution, Lindley distribution, Dagum distribution), and probability distributions Expanded (Marshall-Olkin Exponential distribution, Marshall-Olkin Lindley distribution, as well as the Marshall-Olkin Dagum distribution that was built by the researcher) For the purpose of testing the survival function behavior and comparing the Maximum Likelihood method and Jackknife method to choose the best method for estimating the survival function of the aforementioned probability distributions based on the statistical scale mean integral error squares (IMSE).
Through the criteria (AIC, AICC), it was found that the proposed distribution (Marshall-Olkin Dagum) is more suitable for representing and describing the data of the sample under study.
