<![CDATA[Transaction data is one of the assets in the digital era that can be used to analyze consumer behavior through the application of data mining algorithms, such as association rules. Not all association rules are safe to publish because some contain sensitive information that can impact customer privacy or a company's business strategy. This study aims to implement the Fast Hiding Sensitive Association Rule (FHSAR) algorithm with an item suppression approach that removes certain items with high frequency and supports sensitive rules so that these rules no longer meet the minimum support and confidence values. Sensitive association rules are explicitly determined by the user based on a combination of items considered confidential and predetermined support and confidence values. The dataset used is a public dataset from SPMF. The algorithm implementation is carried out in the form of a Python and Flask-based web application to upload datasets, set threshold values, view suppression results, and display item support visualizations using Charts.js. Evaluation of the algorithm's effectiveness is carried out using four metrics: Misses Cost (MC), Artificial Rules (AR), Item Loss (IL), and Data Dissimilarity (DD). The test results show that the FHSAR algorithm is able to effectively hide sensitive rules with MC values of 1.4%, AR of 0%, IL of 0.9%, and DD of 0.04%. The data sanitization process has minimal impact on the structure and quality of the data, so that the resulting dataset is still suitable for further analysis.]]>
Copyrights © 2026
