A good and correct survey business process is to select a representative sample in order to obtain quality data. However, one of the relevant problems in data quality is the presence of missing values. Missing value is found in almost all large-scale data collections. Missing values can cause all sorts of problems. Therefore, it must be addressed. One way to overcome missing values is the imputation method. Hybrid KNNI-GA and missForest are imputation methods that can be used to handle missing values. Hybrid KNNI-GA uses a genetic algorithm to select the optimum k value and requires predictor variables to perform imputation. Meanwhile, missForest forms a model to carry out the imputation process. This study compares the hybrid KNNI-GA and missForest in dealing with missing values in terms of estimator accuracy and computational performance. The simulation results obtained, the KNNI-GA hybrid is better than missForest in terms of estimator accuracy. Meanwhile, missForest's computational performance is more stable than the KNNI-GA hybrid.
Copyrights © 2022