• Birinder Singh Sarao
• Kamaljeet Kaur

Detection of outliers from large datasets is a complicated issue in data mining nowadays. Discovering outliers is very useful in the detection of unrevealed and unpredicted data in various areas like credit cards fraud detection, criminal behaviours, etc. Many outlier detection algorithms like depth-based outlier detection algorithm, K-Means, Distance-based, Iterative K-Means, etc. are already being used in various surveys, research and review articles. When used on small datasets, these algorithms performance is almost the same, but as the size increases from 50 documents to 1000, their efficiency starts varying. A new algorithm is proposed considering the bulkiness of data and complexity of time. Clustering of the data is done in a typical manner to optimize the process of reducing outliers. The proposed approach is an enhanced version of depth based clustering algorithm, which is implemented on real-world data sets and compared with other algorithms based on various parameters.

Big Data, K -Means, Clustering, Depth Based Outlier Detection, Outlier Detection, Cosine Similarity.

"A Clustering Approach to Reduce Outliers in Large Data Sets", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 3, page no.668-674, March-2019, Available :http://www.jetir.org/papers/JETIR1903N89.pdf

"A Clustering Approach to Reduce Outliers in Large Data Sets", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 3, page no. pp668-674, March-2019, Available at : http://www.jetir.org/papers/JETIR1903N89.pdf