Pre-processing methods of data Mining

Data generation, handling and its processing have emerged as the most reliable source of understanding and discovery of new facts, knowledge and products in the world of natural and material sciences. The emergence of the most efficient techniques in statistical or bioinformatics situations has therefore become a routine practice in research and industrial sectors. Under practical conditions, dealing with large datasets, it's likely to have inconsistencies and anomalies of all kinds to prevent to know real outcomes for practical problems. For accurate data mining computer based techniques of data pre-processing offer solutions that help the data under processing to conform normal structures which in turn considerably improve the performance of machine learning algorithms. In this process, accurate determination of outliers, extreme values and filling up gaps poses formidable challenges. Multiple methodologies have therefore been developed to detect these deviated or inconsistent values called outliers. Different data pre-processing techniques discussed in this paper could offer most suitable solutions for handling missing values and outliers in all kinds of large datasets such as electric load and weather datasets.