Disease risk prediction has become an important area of research in the medical field. As big data becomes increasingly common in the biomedical and healthcare communities, more data accurate analyses of medical data aids, early illness detection, patient care, and community services. Feature selection has demonstrated its efficiency in numerous applications by constructing modest and more comprehensive models, improving learning performance and preparing clean and clear data. This research is focuses on two methods to resolve the analyses feature selection difficulties for big data analytics. An Improved Ant Colony Optimization-based Feature Selection (IACO) and ReliefF algorithms are presented for resolving this issue. The reconstruction of missing data prior to incomplete data was accomplished by means of latent factor mode. Therefore, it was not easy to choose the most appropriate features from structured and unstructured data. The comparative techniques were used to find the most effective features selection among big data. The result provides the significant improvement prediction accuracy when compared to the existing ones.

big data, ReliefF, Feature Selection, Data analytics, healthcare