A Survey on Data Preprocessing Techniques in Stream Mining

Mining static data is itself a tedious task, further it adds complexity in mining stream of data arriving at a high velocity. Data emerging from heterogeneous sources might be impure and this impurity acts as a barrier in the accuracy of the mining algorithms results. The quality of data provided as an input to the mining algorithms could improve the quality of results of the mining algorithms. Preprocessing is used to make the data qualitative by applying certain techniques such as feature selection, instance selection, discretization, integration, normalization, cleaning and transformation, and missing value imputation. In this paper different techniques for Data Preprocessing are reviewed with primary intention of reducing the enormous amount of data and improving the quality of data by handling missing values appropriately. We expect that by applying these techniques, improvement in the Data Mining results will be envisioned.