Data Mining for Auditing - Essay Example

Data Mining for Auditing Contents of the Report Introduction Auditing: An Introduction to the Problem Domain The Solution to the Problem Data Mining: An Introduction Integration of Data Mining with Auditing Introduction The transition of the applied information technology from the primitive file processing systems to sophisticated database systems can be tracked to 1960s. The evolution of the relational database systems coupled with development of data modelling, indexing and organisational techniques have led to the massive utilisation of databases and data warehouses in virtually all business transactions. One such area where data plays a key role is auditing. Auditing is a crucial role carried out by all companies as a test of their own procedures and products. To assist auditors, companies deploy massive databases to capture all relevant data from all departments; this can be used by auditors to assess the company's internal control. However, with increased use of databases, comes a new challenge: how to make sense of the abundant data Auditors are overwhelmed with massive collection of data. Omnipresent personal computers, low cost multi-gigabyte disks, ubiquitous electronics and new generation database languages have made it very simple for companies to capture data and save it without any worries of loss of space, time or computing power. This benefit of databases to companies is also the bane to auditors. However, the effective utilisation of one robust technology will bring sense to the chaos generated by databases: Data Mining. Data Mining aims at converting data to sensible information. It intends to extract information from the data repositories in a manner as needed by the auditor. The auditors, with the help of data mining techniques can 'mine' for the relevant information needed to perform their assessment without having to bother about the irrelevant data. This report aims at analysing the benefits brought about by applying data mining technologies to auditing. As a part of the process of analysing the benefits, the paper also presents the technological overview of data mining, the problem faced by auditors and the tools and techniques data mining provides to alleviate the problems. Auditing: An Introduction to the Problem Domain Auditing is commonly defined as the process of accumulating and analysing information to detect the degree of conformance of the information with the pre-set criteria (Arens & Loebbecke, 2000). During its inception, auditing was an activity performed only to check financial compliance with the goals set. However today, it is an activity that is carried rigorously across all the domains of an enterprise. Auditing involves analysing the information from all departments including manufacturing, operations, human resource, finance and other verticals. Generally, companies hire independent auditors from outside the company to ascertain whether the statements of the company are in conformity with the generally accepted accounting principles (GAAP). However auditing is facing very tough challenges. The demise of major companies such as Enron and Anderson are live examples to limn the depth of negative impact that can be brought about by improper auditing. The complexity of business transactions coupled with investor's complex business practises to gain more profits makes the job of an auditor very challenging (Vijayalakshmi, 2003). To ensure that an objective assessment is reached, an auditor must be presented with data at all levels. The company creates huge databases of statements, records and other data that an auditor is expected to analyse. However due to timing and cost constraints, auditors can not examine every detail behind the stacks of records. With massive improvements in technology such as the development of Supply Chain Management Systems and Enterprise Resource Planning applications, the amount of business transactions performed everyday has grown exponentially. Since, in most cases, a company hires an independent auditor, the auditor will not be aware of the business process and culture of the target company. The auditor therefore, will not have any preset parameters in mind for analysis. Analysing the business process of the organisation requires a lot of time and cost. Secondly, in the process of going through all the data, the auditor may miss several of the important aspects. There is a good possibility that, in the process of making sense of all the data, some of the important driving factors may be skipped. These problems suggest that for efficient analysis of audit data, the data must be presented in a more managed way to the auditor. One robust mechanism to make the process of auditing safe simpler and efficient is the effective utilisation of technology. The Solution to the Problem Technology has definitely impacted the process of auditing over the years. Dealing with complex transaction require more than human judgement. Developments of computer assisted auditing tools are on the rise. The power of automation is a great boon to auditors. One such technological domain that promises to be of great value is Data Mining. Data mining automates data manipulation and very quickly presents data as information to the auditor. Data mining tools provide powerful algorithms that can query data in every way possible. In the sections to follow data mining technology and the tools it provides to assist auditing is analysed. Data Mining: An Introduction Data mining refers to extracting or 'mining' knowledge from large amounts of data. Jiawei Han and Micheline Kamber (Han and Kamber, 2001) aptly state that, the term 'Data Mining' is actually a misnomer. Mining gold from rocks is referred to as 'Gold Mining' and not Rock Mining; similarly, Data Mining should have actually been called as 'Knowledge Mining' since it enables one to extract data in form of understandable information. Data mining tools are indispensable to large organisations as they help the users to sift through vast quantities of information looking for valuable patterns in the data. A pattern may be as simple as that 80% of male cigarette buyers also buy beer. Data mining is the process of discovering unsuspected patterns. The data mining tools generally perform fuzzy searches with very minimal guidance from the user. It assumes that the user himself is not very sure of what is needed. Most data mining tools generally use the following tools to perform the fuzzy searches: (Orfali et al, 2007) Associations look for patterns where the presence of something implies the presence of something else. Sequential Patterns look for chronological occurrences. Clusterings look for high level classifications Data mining involves a systematic hierarchical process with several phases where each phase performs a well defined task. According to CRISP-DM, a consortium that attempted to standardize data mining process, data mining methodology is described in terms of four levels (CRISP-DM, 2000): Processes involve the data mining phases that describe the deployment of data mining tools to solve a business problem. Generic Tasks are the various tangible requirements of the business. These need some input from the users. Special Tasks include the tasks of special interest. Process Instances are the outputs generated from the preceding phases. Integration of Data Mining with Auditing Application of data mining techniques to auditing process is relatively new and yet to be consolidated fully. Although this area is relatively new, it is gaining popularity among the auditor's group. The various popular areas of audit process where data mining can be integrated are presented below. Risk Analysis: This is an area which seems, data mining is tailor made for. By using dependency analysis, various financial ratios of a company can be analysed. The level of each risk can then be marked by using risk triggers. Dependency analysis algorithms can sift through the previous year financial statements to help the auditor analyse the current company trends. Audit Program Preparation: Once the risks have been analysed, the auditor must prepare the appropriate audit program for each vertical of the company. Classification and prediction algorithms assist the auditor in preparing the audit program. Controls Identification and Reliance Assessment: System information data of each vertical in the company can be mined with the help of classification algorithms to identify the various controls existing. With the risk analysis output as the backdrop, the controls can be tested for reliance with the help of prediction and classification algorithms. Testing and Results Evaluation: After reliance assessment, the auditor needs to test and evaluate the functioning of the various business controls. Clustering algorithms provide groups of similar transactions that can be tested with the help of outlier algorithms that identify transactions falling outside the specified group. All the results on various samples can be evaluated with the help of classification algorithms and next set of actions can be determined. Similarly many such detailed analyses can be automated with the help of data mining tools and algorithms. The algorithms supported by modern data mining tools help automating databases of almost any type of business process including banks, government organisations, IT companies, Manufacturing industries and many others. Data mining is definitely the technology of choice for it provides the following advantages: Scalability: The new generation data mining tools are designed to be scalable to work in the framework of any business setting. Auditors can use data mining tools to audit a small organisation with a few hundred transactions a day and at the same time audit huge financial institutions with over a million transactions a day. Handle Complex Problems: Data mining tools are developed to handle complex business problems. Auditing requires the data to be mined in very complex forms. For example for forecasting fraudulent financial statements, auditors will need very specific and transient information from the huge database. Data mining algorithms are powerful to handle such requirements. Uncover Unexpected Information: Data mining helps not only to perform the core auditing functions but also acts as a knowledge tool that provides interesting information about the company activities to those concerned. Learning Capacity: Advanced Artificial Intelligence (AI) has made it possible for data mining tools to learn from their previous experiences. This feature is of great importance to an auditor to whom analysis of previous history log is priceless. Although the integration between data mining techniques and audit processes is a relatively new field, this technology is priceless for reducing costs and time pressures in many business applications related to auditing. References Jiawei Han and Micheline Kamber (2001), "Data Mining: Concepts and Techniques", Morgan Kaufmann Publications Robert Orfali, Dan Harkey & Jeri Edwards (2007), "Client Server Survival Guide", Wiley Press, Third Edition Cross Industry Standard Process for Data Mining (CRISP-DM), (2000), "CRISP-DM 1.0 Step-by-Step Data Mining Guide", found at: www.crisp-dm.org Arens, Alvin A. & Loebbecke, James K. (2000), "Auditing: An Integrated Approach", New Jersey: Prentice-Hall Dr. S Vijayalakshmi (2003), "Challenges before Auditing Profession", ICFAI Press Read More