Estimating Missing values in Randomized Complete Block design using EM Algorithm

Abstract

This research explores the application of the Expectation-Maximization (EM) algorithm to address missing data challenges in Randomized Complete Block Designs (RCBD). Traditional imputation methods often introduce biases, prompting the need for a robust solution. The EM algorithm, anchored in a mixed effects model, iteratively estimates missing values, demonstrating its efficacy in a practical RCBD scenario. Through 15 iterations, the algorithm converges to stable parameter estimates. Root Mean Square Error (RMSE) and bias calculations validate the accuracy of imputations. This study establishes the EM algorithm as a powerful tool for enhancing the integrity of analyses in RCBD experiments with missing data.. KEYWORDS :Expectation-Maximization Algorithm, Randomized complete block design (RCBD), Mixed effects model, Root mean square error (RMSE), Machine learning approaches.