Estimating Missing Data Using Neural Network Techniques, Principal Component Analysis and Genetic Algorithms

The common problem of missing data in databases is being dealt with, in recent years, through estimation methods. Auto-associative neural networks combined with genetic algorithms have proved to be a successful approach to missing data imputation. Similarly, two new auto-associative models are developed to be used along with the Genetic Algorithm to estimate missing data and these approaches are compared to a regular auto-associative neural network and Genetic algorithm approach. One method combines three neural networks to form a hybrid auto-associative network, while the other merges Principle Component Analysis and neural networks. The hybrid network and Genetic Algorithm approach proves most accurate, when estimating one missing value, while the PCA and neural network version is more consistent and captures patterns in the data most efficiently, in the chosen application.