Data mining plays an essential role in the financial sector of organizations as it helps in turning data (raw) into highly useful information. In the present research, the study emphasizes the application of data mining technique in the detection of financial fraud. The focus is made on the background of the research, as well as its significance for future researchers. Different techniques and tools are used for eradicating fraud from the financial department of organizations.
Data mining techniques are important for discovering patterns as well as relationships in data for making better decisions. Based on the viewpoint of Herawati (2015), techniques of data mining is beneficial for anticipating and quickly detecting frauds, besides taking immediate actions for minimising costs. In the modern era of globalization, financial frauds are increasing for meeting earnings expectation. In addition to this, it is also noted that financial frauds generally occur in the financial sector of organisations for increasing management compensation. For example, Kmart's accounting scandal of 2002 is widely common as the company has overstated 10% of its profit for attracting a large number of investors (Villanueva, 2015). It is noted that before the accounting scandal, Kmart had more investment and higher stocks.
The above graph shows that after the financial fraud in 2002, the stocks and sales of Kmart dropped, thereby leading the company to bankruptcy. Moreover, the debt of the company increases, besides losing 57,000 employees (Villanueva, 2015). Other financial frauds that are widely noticed among today's organisations include merging of long-term and short-term debt into a single account for improving the perceived organizational liquidity. Over-recording revenue sales, hidden liabilities, unsuitable disclosure and under-recording expenses are different financial frauds that are widely practiced in organisations.
The impact of financial fraud results in misappropriation of the assets by employees. Moreover, fraudulent report of finance is disseminated to stakeholders, public and investors. Therefore, in order to overcome the negative impact of financial fraud, organisations need to detect it within time by using data mining techniques. The sophisticated techniques of data mining are useful for organisations to search million of transactions for spotting patterns as well as detecting fraudulent transactions (Hassani et al. 2016). Different data mining techniques, such as Random Forests, Boosting trees and Classification trees, need to be used by organisations for detecting any fraudulent behaviour in organisational financial department. Hence, the present topic of research will be interesting and highly important for future researchers as it can help them in carrying forward the study and bringing a satisfactory end to it.
The present research will aim to critically analyse the implication of data mining techniques in financial fraud detection.
1.3 Research Question
Q.1. What is the concept of data mining and its technique?
Q.2. What is the concept of financial detection of fraud?
Q.3. What are the challenges that are faced using data mining techniques?
Q.4. What are the core influencing factors of financial fraud?
Q.5. What is the impact of data mining techniques over financial detection of fraud?
2. Research objective and Literature review
a To analyse the concept of data mining and its technique
b To determine the concept of financial fraud detection
c To evaluate the challenges within the data mining technique
d To analyse the core influencing factor of financial fraud
e To determine the impact of data mining technique over financial fraud detection
The article mainly tends to explores the application of data mining technique within the accounting as well as it tends to propose an organisation framework within these application. Moreover, the article highlights that data mining is the application of specific algorithms which is used for extracting a pattern from data. In addition to this, it shows that data mining key focus is mainly leveraging the data assets for deriving the non-financial or financial benefits. In the technologically advanced market, it is very difficult to detect or prevent fraud in organisational activity. Therefore, most of the organisations emphasised on the selection and application of proper data mining techniques to detect the fraud activities within an organisation which impacts negatively on the organisational profit. However, it has been found that there is various kind of fraud activities are exist which directly on the organisational monetary fund. The fraud type mainly includes credit card frauds, medical insurance fraud, and so on. Most of the fraud activities are associated with the credit cards, including a duplicate of signatures, wrong information of address and identity along with skimming (Amani and Fadlalla, 2017). On the other hand, fraud medical claims also become a significant issue for life insurance organisations.
Therefore, the organisations started to adopt various kinds of data mining techniques like regression, linear, SPSS analysis along with peer group analysis and breakpoint analysis. It has been found that such kind of data mining techniques positively on the prevention of financial fraud in organisational function. In the study, the main data sources were the leading accounting journals and in here there are a total of 209 papers satisfied in terms of inclusion criteria. Moreover, they also used the method of article filtering to make sure all the article are related to the topic of data mining application in accounting department. Moreover, the main debate in here is the showcase of data mining application within the accounting sector.
From the article the results mainly denotes that between the year 1989 to 2014 there were a total of 209 application from their 23 in conference proceeding papers and 186 described in the journal articles. It was also found that there was a leap within the modelling sophistically which is following a major worldwide financial crisis of the year 2008. It is also found that financial accounting mainly tends towards examining financial performance as well as analysis. Moreover, it was found that the earliest data mining application was built in term of neutral networks model for forecasting the quarterly accounting earnings.
Data mining generally stands for a process for discovering actionable information from the large sets of data. The data mining mostly utilises the mathematical or technical technique for deriving patterns and trends which exist in the data (Tan, 2018). On the other hand, the data mining process or technique generally associated with the business process of an organisation in the context of mining their financial data. However, data mining of an organisation can be done by following some significant technique for extracting potential data from the bunch of organisational data. The data mining technique mainly includes tracking patterns, clustering, prediction and regression (Lausch et al. 2015). In respect of the research topic, the data mining techniques are mainly liable for the collection of transactional details and financial data of the Australian organisations.
Hence, in this article the main methodology used in terms of data sets which were available within the UCI Machine Learning Repository and it is applied in different visualising technique. Hence, this analysis shows that it is relatively highly developed and it is accurate if it is widely available. Moreover, in here the analysis is done in respect of basic concepts and algorithms, advanced concepts as well as basic concepts. Apart from this, the algorithms and additional issue analysis is also made for finding the results. Apart from this, there can be showcase of the grounded theory for system methodology associated with analysis a data collection. The grounded theory also helps to analyse the hypothesis by analysing the collected data through interview or survey from the market (Zamawe, 2015). Moreover, this theory helps to analyse Table 2.1: 4 C concept of data
Based on the above-mentioned steps, it can be analysed that by analysing the theory, the research will be able to analyse the collected data from the respondent regarding data mining and financial fraud. Moreover, it also helps in analysing the hypothesis based on the impact of data mining for detecting financial fraud within organisations.
Business's accelerated globalisation has simply tends towards weakened regulatory capacity of law towards the paid attention of fraud detection within the recent years. It has been found that the peer group analysis helps an organisation to monitor and understand customer behaviour with respect to a certain time. The organisation can create peer group of account for the every credit cardholders which helps in measurement of same behaviour exhibition. Moreover, it helps in all the customers by dividing their peer groups with respect to similar behaviours (Van Vlasselaer et al. 2015). In this technique, if any of the account from peer groups performs subsequent behaviour that strongly deviates can be considered as the fraudulent. Therefore, it can be analysed that this data mining technique impacts positively on the detection of fraudulent. On the contrary, if the peer group account is hacked by an individual, it will be impossible to detect the anonymous behaviour of any individual which may fail to detect the fraudulent.
Another technique is called "Breakpoint analysis" which helps in distinguishing the spending activities which is supported by the transaction information within a single account. It impacts positively on the matching of new transactions with the prior spending activities for sporting features. Therefore, it can be analysed that this data mining technique positively on detecting financial fraud detection. On the contrary, if an individual put any wrong inputs regarding financial activities can impacts negatively on the identification of financial faults due to the improper balancing of the activities (Liu et al. 2015). Moreover, it also has been found that most of the significant data mining models help organisations to find out the cost frauds at the time of saving adjusters time. Based on some data mining techniques, it can be analysed that it provides a different kind of benefits to the organisation. However, it has been found that the banks and financial institutions are profited by data mining in the context of identifying probable defaulters. It helps the organisations to issue credit cards for the customers who are based on past transaction records and user behaviour.
The latent class model generally associated with the multivariate variables for setting latent variables; besides, it also can be recognised as the latent variable models. As the latent variable is discrete is able to characterise the pattern of conditional probabilities, which indicate the chance which the variables can take on certain values (Bakk and Vermunt, 2016). Moreover, the model also can be recognised as structural equation modelling, which helps to find groups within the multivariate categorical data. Besides the latent class model also helps in identifying dependent and independent variable both. Therefore, it can understand that by utilising this model, the researcher can find out the independent and dependent variable both. It helps the researchers to understand what kind of factors can influence financial fraud detection.
In this article it can be seen there is the use of Ballpark Matrix of Data along with the measurement method estimation. Apart from this method there was also used the CSMAR Data which includes the financial frauds aspects. The analysis highlights the introduction of Random Forest model for financial fraud in terms of data mining. Hence, to detect the financial it is found that the use of Random forest method has the highest accuracy regarding detecting frauds. The main debate focused within this article is where Random forest method provided the correct fraud detection or not (Liu et al. 2015).
Financial fraud can be recognised as the international deceptive act associated with the financial transaction along with personal profit-making. As per the business context, financial fraud mainly stands for the loss of monetary fund of an organisation due to some illegal activity of the employees and customers. It has been found if an organisation did not track the financial record of each and every transaction related to the business can face the financial challenges or difficulties (Mansor, 2015). It can impact negatively on the market position of any particular organisation or organisational profit. Besides, mostly that organisation who deals with their customer through online transactions suffers more than traditional organisations (Isukul and Chizea, 2015).
Moreover, the frauding of financial data can be successful if the data mining technique of an organisation was not strong. Therefore, the financial report of the particular organisation would be affected due to loss of financial records. Besides, the poor data mining technique was unable to detect the small transaction data from a bulk number of organisational data (Baxter, 2016). In the research study, most of the Australian organisations were facing financial fraud cases because of having improper data mining technique. Therefore, it aims to analyse the impact of data mining
The main methodology used in here is the secondary sources by suing the journals for getting the information's regarding fraud. In addition to this, the discussion also shows the usage of two theories which contributes for understanding fraud especially through forensic accountants and fraud examiners. The main highlighted debate is between the similarities between FDT and FTT (Mansor, 2015).
Fraud is mainly debilitating factor which affects the growth of business and continues towards a major problems for organisation. In addition to this, the article showcases that fraud schemes are seems to be highly sophisticated as well complex with the implementation of technological advancement. Moreover, the financial concept can be analysed with the three statement model concept which analyse the financial factor. Moreover, it also has been found that most of the gaming industries of the various countries including Australia faced lots of financial frauding due to having poor standardized online identity verification and authentication tool (Mangala and Kumari, 2015).
Based on the above-mentioned picture, it has understood that the three statement model helps to link among three factors of financial accounting. The model mainly emphasised on the linking among income statement, cash flow and balance sheet one connection (Guay et al. 2016). Therefore, it can be understood that by implementing this model, an organisation can analyse their financial data by seeing the information of three accounting statement together. It impacts positively on customer data tracking and financial report making. The main methodology used within this article is the use of different theories to showcase the fraud within the financial sector. The main debate in here highlights the combat fraud which tends to make the business understand regarding the elements and risk factors.
The article mainly highlights the big data where new technology paradigm regarding the data which are generated at a high level volume and velocity. The article integrated towards the view of data mining regarding within the past 20 years. However, in the context of mining data, some criteria must be matched with the data mining techniques; otherwise, it may create challenges to mine data such as
It has been found that more collected data were incomplete and noisy to understand. Therefore, the validity and reliability of the particular data were not potential for mining. On the other hand, the spelling mistakes at the time of putting data impacts negatively on the finding of real mail IDs of the customers (Jin et al. 2015). Therefore, it can be understood that it can create challenges for mining previous customer data records.
The real-world data was usually stored on various platforms in the context of distributed computing environments. Therefore, it will be very difficult to find or bring entire data to the centralised data respiratory for an organisational and technological reason. As an example, all the different regional organisations utilise or their local server to access customer data (Van Der Aalst, 2016). Therefore, it will be very difficult for an organisation to centralise all customer data into one server.
It has been found that every country has different activities or rules related to the data protection act of consumers. Moreover, the different country holds a different policy for utilising customer data (Bello-Orgaz et al. 2016). Therefore, it will create a great problem for a globalised organisation to utilise the data of the customer transaction of their different branch located in a different country (refer to appendix 1).
The main methodology used in here is the lexical based method along with the machine learning method for the sentiment analysis. These were used in terms of showcasing the data mining techniques. The main debate in this article highlights the impacts and challenges faced by the big data in terms of data mining.