<![CDATA[Survival analysis is a statistical method used to analyze the time until a certain event occurs, such as death, equipment failure, or a patient's recovery. The parametric approach in survival analysis assumes that the survival time data follows a given probability distribution, so the selection of the right distribution greatly determines the model's accuracy. This study aims to compare the performance of three commonly used parametric distributions, namely Weibull, Exponential, and Log-Normal distributions, on survival data using goodness-of-fit criteria. Parameter estimation was carried out through the Maximum Likelihood Estimation (MLE) method, while the model evaluation used Log-Likelihood, Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) values. The data used were simulation data generated from the Weibull distribution with the form parameter k = 1.5 and the scale parameter λ = 0.05, with n = 100 observations and 20% censored data. The results show that the Weibull distribution produces the smallest AIC = 214.73 and BIC = 219.54 values compared to the Exponential (AIC = 238.61, BIC = 240.81) and Log-Normal (AIC = 218.45, BIC = 223.26) distributions, so the Weibull distribution is chosen as the best model. These findings confirm that the flexibility of the shape parameters on the Weibull distribution provides an advantage in modeling survival data with non-constant hazards]]>
Copyrights © 2026
