Machine learning algorithms have the potential to add flexibility and adaptability to current data examination methods, which mainly consist of specific rule based queries. Although machine learning is a methodology that is widely applied in many scientific domains, it is a topic which is rarely researched in financial sectors. Financial transactions are inspected by auditors and their aim is to create the company's financial statement at the end of the year. However, these inspections are relying on predefined knowledge and do not reflect the whole dataset but rather on specific subsets. In this research, we proposed an anomaly detection model by a machine learning approach to support auditors in finding anomalous behavior out of the whole dataset. This investigation used apriori, which is an association rule based and unsupervised machine learning algorithm that can identify unexpected account combination in the dataset. For this analysis, we take real datasets that consist of transaction details from Bearo of Finance and Economic Development Financial system recorded in 2017. R machine learning tool have been used to do the analysis on the data and make the anomaly detection model. The Model designed by this tool makes anomaly detection available to the auditors. This model is able to visualize the whole data set and detect account pairs which have an unexpected amount of money compared to their frequent behavior. Then, evaluation has been done for the final model result by the domain experts and using synthetic dataset. Finally, this research results a tool for auditors that would increase the quality of review of the financial transactions and, in hence, would support them in the identification of anomalous financial transactions.