Class center-based firefly algorithm for handling missing data

Abstract A significant advancement that occurs during the data cleaning stage is estimating missing data. Studies have shown improper handling leads to inaccurate analysis. Furthermore, most studies indicate occurrence of irrespective correlation between attributes. However, an adaptive search procedure helps determine estimates when correlations attributes are considered in process. Firefly Algorithm (FA) implements imputation by determining estimated value closest others' value. Therefore, this study proposes a class center-based approach model for retrieving considering attribute process (C3-FA). The result showed firefly algorithm efficient technique obtaining actual with Pearson coefficient ( r ) and root mean squared error (RMSE) close 1 0, respectively. In addition, proposed method has ability maintain true distribution values. This indicated Kolmogorov–Smirnov test, which stated D KS dataset generally closer 0. accuracy evaluation results using three classifiers produces good accuracy.