Usage of Business Intelligence in Banking Industry

With the improvements of digital technology, banking industry gains huge benefits. It provides mechanism to data warehousing by storing data in the branches and increases the number of access points to the bank accounts. Banking system becomes steady in technically and customer oriented with transactions which is concluded in the online, ATM cash and cheque deposit machines and electronic wire transfers. All the transactions and the related data have been stored. Thus, nowadays banks maintain large electronic data warehouses as the bank electronic storage. Constantly data is grown up in terms of size and dimensionality. By using the help of big data analytic techniques those huge data will be turns into the most beneficial asset of the banks. Those data consist of interesting patterns and beneficial knowledge. As a result of that, banking industry has huge possibility to applying data mining techniques to identify such patterns and knowledge individually to help critical decision making processes such as managing risk, marketing, money laundering and fraud detection. This article is about the How banking industry uses Data Mining techniques to detect frauds efficiently.

Annually banks lose millions of dollars due to the various frauds happening in the banks. By detecting those frauds banks can be reduced the damages or prevent from them. Fraud detection can be known as the process of recognizing frauds separately from the trustworthy actions or transactions. In other words fraud detection is the process which divides all the operations into two classes such as legitimate and fraudulent. Credit card transactions and financial statements investigation can be known as most significant areas in the banking industry which can takes the benefits of fraud detection. Banks makes the decisions about the credits by using the financial statements provide by the customers. Sometimes those provided statements may consist of overstated profits, sales and assets or It may contain understate their liabilities and losses. Those statements may have been review, but frauds that I have mentioned above are not easy to detect by using normal examine procedures.

Most of the banks starts to using data mining techniques recognize legitimates and fraudulent. Fraud detection is one of the most important and popular area that data mining can be used to gain the benefits. Identification process of fraudulent actions is a rising issue for many banks in the industry and most banks takes help of data mining techniques to detect and report more fraudulent actions. Two particular mechanisms have been created by financial institutions to detect and recognize frauds. First one is they uses the data warehouse which is maintains by third party and identifies particular information’s regarding the frauds by using data mining application. As a result of that, bank can cross reference identified information with their database to detect signs of internal trouble. Next approach is identification of fraud information is strictly based on banks internal information. Most of the banks in the industry use both approaches partially as hybrid. presently, “Falcons fraud assessment” can be known as a system which is success in detecting frauds. It is being used by nine of the top ten popular banks in the industry. The data mining application also help for the banking industry to focus on the procedures of analyzing data of customers to identify information about behaviors which have possibility for lead to frauds.

With the advancement of banking services, Banking industry had to face lots of losses due to the various frauds happening in the bank. According to the annual reports, credit card based transaction frauds (online/ATM) and financial statement based frauds can be known as most significant areas which leads to lose millions of dollars annually (before implementation of proper business intelligence solution). As a solution, Banks uses business intelligence applications such as data warehousing, big data analytics and data mining effectively in a fraud detection process to reduce those frauds and losses.

Bank properly maintains and organizes data using data warehousing application. Those organized data is being analyze and identifies spending patterns, credit information’s, behaviors and other relevant information and grouping using big data analytic techniques. On other hand, by using data mining techniques, they understands customer in terms of behavior, choice of investments and customer demographics individually from big data. This knowledge of customers is used to help customers and get better returns. Moreover banks uses that information’s to making better decisions on frauds detection, customer relationship management and so on. Most of the financials have been improved their fraud detection, fraud case handling efficiencies and false positive rates by implementing proper data mining techniques. Top 10 banks appropriately maintains data on frauds of area of operation under which frauds have been perpetrated. According to those data, banks has been reported two categories under the frauds such as transaction based frauds and financial statement based frauds.

Sometimes stored transactions history and demographics of the customers are providing information to defraud the bank. Techniques of data mining application help banks to analyze those transactions individually and identify patterns which leads to fraud. Banks more focused on fraud detection. Thus, it is significant to find which transactions are not the transactions that the user would be doing. Therefore, there should identify which transactions are not suitable into specific category or which transactions are not fit with standard and it is also necessary to determine which of the user’s actions correspond to their natural behavior and which are not correspond to their natural behavior automatically and intelligently. With the help of proper data mining techniques, They detects doubtful activities within the data in a formal way. System examine the transaction history of the customer to identify previous locations which is customer has been completed their transactions by using the cluster analysis technique in the data mining and compare those locations with current transaction location using anomaly detection technique. If the current transaction location does not similar with their previous transactions, it will be noted as suspicious transaction. Not only had that, but also at the same time it examines patterns of the user’s behavior by organizing related transactions together using cluster analysis technique. In order to inspect abnormal transactions, current transactions are compared with the identified user’s common behavior patterns using anomaly detection technique. Furthermore, if it is cyber credit card transaction, system will investigate significantly the regular merchant websites which is card owner regularly makes the payments for services or goods, the geographical information where goods have been shipped to by the owner, email address and the phone number which uses regularly for their transactions. Current transactions which are not corresponding with the any of these identified patterns and owners’ standard, it will be automatically and intelligently identified as a distrustful transaction and steps have been taken accordingly. This is the ordinary procedure of detecting transactions based frauds in the bank using the business intelligence applications.

Besides of transaction based fraud detection, banking industry uses artificial neural network models in data mining to detect financial statement based frauds. Those models have been created based on set of patterns and significant information’s which is analyzed by using the clustering analysis technique. Those models have two classification methods. In the step one, the model has been trained using sample data. Those data is arranged using attributes and tuples by placing properly the attributes which is belong to the class label and every attribute falls under every tuple. This one also identifies as supervised learning. In the step two, performance of the current model is found by classifying the object which is not present in the training and validation set. The input values that uses in the algorithm are composed of ratios will be derived from the statements of financial such as income statements and balance sheets to predict the accuracy of classification method. As above, Banks has been creates separate data sets known as legitimate set and fraudulent set. The statements provides by the customer is being compared with the created neural network models using anomaly detection technique and identifies which statements are fraud and which are not. Further those statements should be annual statements and each value is being compared with the values of assets, profits and other related information intelligently taking it as percentage by using mathematical calculations. In addition to this, it is also comparing statements by yearly to identify the frauds. Moreover, in the fraudulent financial statements supposedly had more activation languages, less usage of lexical diversity and group reference as compared to a non-fraudulent one. By using those information’s banks identifies fraud financial statements and decisions is being taken accordingly.

Let’s check on their main techniques and appropriateness

Data Warehousing

Banks uses data warehousing technique to organize and maintain data properly to use this data in a various sectors.

Big data analytics

Stakeholders uses big data analytic technique to analyze and identify credit information, spending patterns, social media information and other relevant information’s to use in marketing, customer relationship management and so on. This technique is more efficient in analyzing complicated and huge data.

Data Mining

Banks uses data mining techniques to identify specific pattern and behavior of individual customer from the big data and detect frauds through the pattern. As an illustration, this will analyze particular pattern and behavior of customer separately and compare it with the current transaction. For this process, it uses techniques such as,

Cluster Analysis

Banking industry uses the clustering technique to analyze data and grouping accordingly. Clustering technique helps to organize the data into related clusters which helps to retrieve uncomplicated data. Cluster analysis technique is uses for separating data into related components to visible patterns and order. This technique is appropriate because it can be used to create neural network models or directly in identification process (anomaly detection).

Artificial Neural Network

Bank uses neural network to create specific models using set of patterns and information which is identified in the cluster analysis process. It will create two data sets (models) such as legitimate and fraudulent set to use in a detection process. This technique used with anomaly detection technique to detect financial statement based frauds

Classification

They uses classification technique to classify neural network models by using sample data. This technique is appropriate because neural network result accuracy based on the classification set (training set) .better classification leads to better result.

Anomaly Detection

Partners of financial industry uses anomaly detection technique to compare both transactions and financial statements using created models and identified patterns. It is being used with both neural network models and cluster analysis patterns. It is more efficient and higher accuracy level in comparing and detecting.

Those data mining techniques which are used by the banking industry is appropriate and suitable due to the after implementation they successfully reduced lots of possibilities of having frauds in the bank and both processes result has more than 90% of accuracy.

There are some challenges and issues in those techniques as well that they still struggling to overcome. Few of them are follows.

• There are some circumstances; some of the transaction which is done by the real customer is being noted as suspicious one due to the changes of their behaviors and In the cyber credit card transactions, there are possibilities of having incomplete transactions due to the changes of shipping address, contact number with the changes of region.
• Millions of transactions have been processed every day in the banks. As a result of that, it requires highly efficient data mining techniques to extract huge amount of data from their database.
• The data labels which are related to frauds are not immediately available. Usually most of frauds aware later, may be they already happen. Constantly user’s behaviors are changing. Thus it is not easy to track their behaviors directly.

Conclusion

Regularly banks in the industry stores and maintains millions of data each day. With the improvements of digital technology, it increases the possibility of having data warehouse to organize and maintains those data. Moreover, those data consist of specific patterns and information’s which can be used in the decision making processes. Many banks in the industry use big data analytic to identify and group those patterns and information’s to use in a different purposes such as social media marketing, customer relationship maintains and campaigns. On other hand, by using business intelligence application such as data mining, those data is being analyzed and identifies particular patterns and knowledge individually from the big data and use it in critical decision making processes. Top most popular and largest banks are uses BI application to improve their businesses and strategies. Particularly uses BI application to detect transaction based frauds and financial statement based frauds. Not only that, but also it uses BI applications in customer relationship management, marketing and money laundering application areas. Further, it can say those business intelligence applications have relation between each other and properly use of those relation leads to gain lots of benefits for banking sectors and other organization’s as well. Those BI applications provide various useful techniques to conduct their businesses effectively and efficiently.