Abstract:Data mining is one of the most essential fields which is growing with rapid in technologies of information technology and data mining is also becoming purposefully important area for many business companies including banking and financial service industries. Data mining the is a process of evaluating the data from various point of view and summing it into useful information. Data mining help the banks and financial service to identify the uncommon pattern i.e hidden pattern in group of customers and notice unknown connection in the data.
Now-days, customers have so many options to their business wherever they can do. Before data investigation methods were known for separating quantitative and measurable data qualities. These systems encourage helpful data identifications for the managing a banking sector to avoid customer attrition. Customer maintenance is that the most imperative issue to be analyzed in the present in different modest business scenarios. We will look all kinds of data mining tools and technique which can be used to overcome issues due to lack of usage of data mining technologies.
Introduction:Banking and financial services industry face major problem in terms of fraud. The main problem is that it’s difficult to detect or identify or prevent the frauds in this industry and to catch the fraudulent is also difficult task as they always use sophisticated tools and technologies to lure customers like using spamming, social engineering and many more. Let’s put some light on some of the data mining technologies and its implication in Banking and financial services industry like customer retention, digital marketing, Risk management and fraud prevention and detection.Technologies of Data Mining: Utilizing data mining system, it is easy to assemble an effective prescient model and envision the report into significant data to the customer. The following figure shows the stream of data mining method in our system Model.Figure AAs you see on the figure A data mining techniques and algorithm which are being used in Banking and financial services industry .The Customer retention is one of the main purpose of any industry especially in this sector ,In customer retention we use supervised learning method and then we implemented Decision Tree by using CART algorithm .We all know that prevention is better than cure likewise preventing the fraud is better that detecting the fraudulent banking after its occurrence. Decision tree, Support Vector Machine (SVM) and logistic Regression are also used for approving the credit card request. For Fraud detection purpose Clustering and EM algorithm are used.Customer Retention in Banking and financial services:Today, customers have numerous opinions with relation to where they can choose to do their business. Executives in the banking system, therefore, must be aware that if they are not giving every customer their full attention, the customer can simply find another bank that will analysis techniques were headed toward extracting quantitative and applied data characteristics. As we are using supervised learning which help in learning in direct way, it’s directed by previously known depended attribute or predictors.To enhance customer maintenance, three stages are required: 1) Measurement of customer maintenance; 2) Identification of underlying drivers of abandonment and related key administration issues; and 3) The advancement of remedial activity to enhance maintenance.A) Decision Tree(Burez and Van den Poel, 2007) Decision trees are the most popular predictive models. (Burez and Van den Poel, 2007) A decision tree is a tree like graph representing the relationships between a set of variables. Decision tree models are used to solve sorting and prediction problems where instances are classified into one of two classes, typically positive and negative. Decision tree models are determined in a top-down manner. It involves two phases: Tree pruning and Tree building : It starts from the root node that represents a feature of the cases that need to be classified. Following a similar method of knowledge gain analysis, the lower level nodes are constructed by mimicking the divide and conquer strategy. Building a decision tree incorporates three key elements:1- Identifying roles at the node for splitting data according to its value on one variable or feature. 2- Identifying a stopping rule for deciding when a sub-tree is created.3- Identifying a class outcome for each terminal leaf node.Decision tree models square measure accustomed solve classification and prediction problems where instances are classified into one of 2 categories, usually positive and negative, or churner and non-churner in the churn classification case. These models are represented and evaluated in a top-down manner. Developing decision trees involves two phases: Tree building and tree pruning or To perform the logarithmic function can be applied to obtain the logistic function. Logistic regression is simple, easy to implement, and provide sensible performance on a good type of issues (Michal Meltzer).Fraud Detection in Banking and financial services systemIn fraud detection case we will use clustering and Estimation method Algorithm .(I) The Clustering Model:In this model it clusters same type of data by grouping in order to help in simple retrieval of data. This analyzing will help that to cluster in the same group from that we can identify order and patterns can be evident.TO know about this cluster region we need to find highest value i.e Maximum difference (DIFFmax) and then (DIFFmax) will furthure splits in to NInterval segments.Ninterevel , will have the attribute value called Npoints. N Interval can be found using another way of looking. Such calculation of Ninterevel is based on the assumption that a twofold increase of Npoints will be equal to Ninterevel plus one. Mukhanov, Lev. (2008). For each found segment the calculation of the average value.The final result of classification results for each parameter: Result =w1 x Class1+ w2 x Class2 +… + wn x ClassnEM AlgorithmWe use Gaussain mixture model is used which is the sum weighted component densities of Gaussian form. (Onukwugha, Chinwe. (2018).)The p(x”‚j) is the jth component density of Gaussian form and the P(j)is its mixing proportion. This methodology specialize the final model by re-estimating the blending proportions for every user dynamically once every sampling amount as new knowledge becomes out there. Whereas the suggests that and therefore the variances of the user specific models square measure common, solely the blending proportions square measure totally different between the users’ models. It is necessary to retrieve the information from the last k days and adapt the mixing proportions to maximise the probability of past behavior. however this approach needs an excessive amount of interaction with the request system to be employed in apply. To avoid this burdensome process of knowledge, this methodology formulates the partial estimation procedure victimisation on-line estimation. The on-line version of the EM formula was 1st introduced by Nowlan.Digital Marketing:Digital Marketing is one the core part of banking industry as everyone goes through e-transactions while buying online or file meaning customer uses credit cards and bank accounts for transaction purpose,Bank experts can likewise investigate the past patterns, decide the present interest and figure the customer conduct of different items and administrations so as to get more business openings and foresee conduct designs. Data mining system likewise recognizes gainful customers from non-productive ones. Another real zone of improvement in keeping money is Cross moving i.e. banks make an alluring offer to its customer by requesting that they should purchase extra item or Services.Risk Management:This area has to take care in effective way why because by taking risk on any services of bank industry either you may loose the customer or you gain it, likewise, Data mining procedure recognizes borrowers who reimburse credits regularly from the individuals who don’t. It also predicts when the borrower is at default, regardless of whether giving credit to a specific customer will result in bad loans. Bank administrators by utilizing Data mining system can likewise break down the conduct and dependability of the customers while moving charge cards as well. It additionally serves break down whether the customer will make provoke or delay installment if the credit cards are sold to them.Conclusion:Data mining is a technique used to abstract vibrant information from current huge amount of data and enable improved decision-making for the banking and financial services industries. They use data warehousing to coupled various data from databases into an suitable format so that the data can be mined. The data is then analyzed and therefore the data that’s captured is employed throughout the organization to support decision-making. Data Mining techniques area unit terribly helpful to the banking sector for higher targeting and getting new customers, most valuable customer retention, automatic credit approval which is used for fraud prevention, fraud detection in real time, providing section primarily based merchandise, analysis of the shoppers, dealing patterns over time for higher retention and relationship, risk management.