Morad, N. (2023). Estimation of Gamma Distribution Parameters with Incomplete Data. The Egyptian Statistical Journal, 67(2), 50-62. doi: 10.21608/esju.2023.232299.1018
Naglaa Abdelmomeim Morad. "Estimation of Gamma Distribution Parameters with Incomplete Data". The Egyptian Statistical Journal, 67, 2, 2023, 50-62. doi: 10.21608/esju.2023.232299.1018
Morad, N. (2023). 'Estimation of Gamma Distribution Parameters with Incomplete Data', The Egyptian Statistical Journal, 67(2), pp. 50-62. doi: 10.21608/esju.2023.232299.1018
Morad, N. Estimation of Gamma Distribution Parameters with Incomplete Data. The Egyptian Statistical Journal, 2023; 67(2): 50-62. doi: 10.21608/esju.2023.232299.1018
Estimation of Gamma Distribution Parameters with Incomplete Data
Department of Applied Statistics and Econometrics, Faculty of Graduate Studies for Statistical Research, 12613, Cairo University, Egypt
Abstract
The missing data problem has been broadly studied in the last few decades. Some researchers studied estimation and hypothesis testing for different distributions .The contributions of the current study was to obtain the estimators parameters of the gamma distribution with missing data for one and two populations. The estimators are obtained using the maximum likelihood method. To compare the suggested maximum likelihood estimators with estimates that might come from various estimation techniques, including the listwise method and the mean imputation method, a simulation study has been conducted. A simulation study with three distinct percentages of missing values in the data sets: 10%, 20%, and 30% as well as three different sample sizes (10, 30, and 50). The estimators' criteria's mean square error (MSE) and the relative absolute biases (RAB) were utilized for comparison. The results demonstrated that the use of mean imputation or the maximum likelihood method for the scale parameter (θ) but the listwise method for the shape parameter (k) is preferred as the percentage of missing data increases. Use the mean imputation approach for the scale parameter (θ) as the data set size grows.
View on SCiNiTO
Top