Outlier Detection of Transaction Data Using DBSCAN Algorithm
The supermarket is one means of marketing the company's products. Marketing activities undertaken with supermarket provides a wide range of types of products from different companies (as producers). Consumers prefer to go to the supermarket than traditional markets due to promo. For example, the products offered were given discounted half price of normal price. Consumers tend to buy more of their needs so that existing stock items in the supermarket can be drastically reduced. Therefore, the supermarket had to anticipate in order to not shortage of stock in the warehouse. Various techniques in data mining can be used, one that is an outlier detection. The role of an outlier detection is needed in order to detect abnormal transactions including candidate anomalies and normal transactions and will be help the supermarket in anticipation of running out of stock items. Outlier detection is an outlier search process on dataset and is one of the first steps to be able to perform analysis of data coherent. The main objective in outlier detection is to detect data with properties/state data with different data or are most of the anomalies found in multidimensional datasets. One of the formidable algorithms for detecting outlier is DBSCAN. Therefore, in this study, the author will use the technique of outlier detection algorithm with expected DBSCAN can help supermarket in anticipation of running out of stock items. The result from research that has been done by calculating 1862 products there was no product data that classified as outlier, whereas by calculating 100 first products there are 4 products data that classified as outlier, products with id 80069449, 80015728, 82024920, 80021527.