<![CDATA[With relatively small transaction value but frequent, manual supervision of goods purchasing for operational needs such as inventory purchasing (account code 521811) becomes very expensive and time-consuming. As a result, spending on this 521811 account becomes vulnerable to fraudulent practices. With the highest realization purchasing value for account 521811, the Budget Section (BA) 060-Polri faces the most significant risk of fraud on that account compared to others. Machine Learning Modeling can be used to automate internal control over this risk with much greater efficiency and effectiveness. Usually, one of the early indications of fraudulent transactions is the irregularity of the transaction pattern from normal behavior (anomalies). This study tries to detect anomalies using the Isolation Forest method at BA 060. The experiment results show that the model is able to detect anomalies of inventory purchasing (account 521811) in working units of BA 060-Polri optimally at the contamination parameter value of 0.3%. Further development of this model can be used as an additional feature in the payment module of the SAKTI application to automate anomaly detection when the operator performs the transaction inputting process.]]>
Copyrights © 2023
