Comparison and Evaluation for Grouping of Null Data in Database Based on K-Means and Genetic Algorithm
- K. Pandole, N. Bhargava
In real life, users of most of the database systems face problems related to missing values, whether these values are unknown to the users or inapplicable. The Database is an image of the real life and life is always incomplete. For this reason many researchers tried to complete the database information through estimating the missing values by applying several algorithms to gain high estimated accuracy rates. This estimation is through predicting and replacing the missing values with approximate values obtained from the development of an algorithm that combines two data mining algorithms (Decision Tree and K-nearest neighbor) for estimations and predictions. The treatment also checks and validates the stored and estimated values through comparing them with internally set business rules. Using Oracle DBMS, a framework is designed and implemented for the estimation of Null Value utilizing the standard database “ADULT”, consisting of (32561) tuples, obtained from UCI Data Repository.