Implementation of Business Intelligence on Banking, Retail, and Educational Industry

Information technology is useful to automate business process involving considerable data transaction in the daily basis. Currently, companies have to tackle large data transaction which is difficult to be handled manually. It is very difficult for a person to manually extract useful information from a large data set despite of the fact that the information may be useful in decision-making process. This article studied and explored the implementation of business intelligence in banking, retail, and educational industries. The article begins with the exposition of business intelligence role in the industries; is followed by an illustration of business intelligence in the industries and finalized with the implication of business intelligence implementation.


INTRODUCTION
*OREDOL]DWLRQ HUD KDV FKDQJHG SHRSOH LQ WKH industrial world in managing their business. Various products are introduced to people from various segments to attract attention so that in the end they will use the products. The use of information technology in automating business processes in information system where a company will save transactional data in a large amount. The use of information technology allows companies to automate business process. In the big data era, companies have to manage large GDWD WUDQVDFWLRQ WKDW LV GLI¿FXOW IRU D KXPDQ EHLQJ WR DQDO\]H GLUHFWO\ DQG GHGXFH XVHIXO LQIRUPDWLRQ IURP the data [1].
Data mining can give contribution towards the solution of business problems in the industry by identifying the current pattern and trend, how the behavior of stage funds towards the condition of the economy, politics, and social. Correlation between various variables in business data cannot be directly seen by the managers since the volume of data is overly large. Managers require additional works to reach a conclusion regarding the behavioral pattern of the customers. Also, many additional works are required to comprehend, dissociate, preserve, and nurture favorable customers. Business intelligence and data mining help the managers and products' managers identify various classes of customers and develop compatible products or services with the customers' needs and or the act of determining pricing strategy to obtain better revenue management [2].

METHOD
Research studied and explored the implementation of business intelligence in banking, retail, and educational industries and its effects in increasing the decision-making process to solve the industry business problems. Article begins with an exposition regarding the role of business intelligence in each industry. It illustrates and discusses the implications of business intelligence.

'H¿QLWLRQ RI %XVLQHVV ,QWHOOLJHQFH
%XVLQHVV LQWHOOLJHQFH LV GH¿QHG DV D SURFHVV RI H[WUDFWLQJ WUDQVIRUPLQJ PDQDJLQJ DQG DQDO\]LQJ business data to support decision making process [3]. The process involves a set of data obtaining from a data warehouse. The process of business intelligence LQFOXGHV ¿YH VWDJHV Data sourcing; the system of business intelligence can extract data from various data sources and various business units such as marketing, SURGXFWLRQ KXPDQ UHVRXUFHV DQG ¿QDQFLDO 7KH extracted data should be cleared, transformed, and integrated for analysis. Data analysis; in this stage, the data are converted to information or knowledge through YDULRXV DQDO\VLV WHFKQLTXHV VXFK DV YLVXDOL]DWLRQ DQG data mining. The result of the analysis process can help the management understand the situation and make a better decision. Situation awareness; the awareness towards situation can provide deeper comprehension in current situational decision based on the result of data analysis.
Risk assessment; the awareness towards VXI¿FLHQW YDULRXV VLWXDWLRQV FDQ KHOS PDQDJHUV SUHGLFW the future, identify threats and chances, and respond in accordance with needs. Nowadays, business operates in complex environmental condition. The decision making of business is more likely to be accompanied with risks that are coming from external and internal environments. Hence, it can be concluded that risk estimation is a main function in business intelligence system. Decision support; the main purpose of business intelligence is to help managers in making decision wisely based on the current business data.

7KH $UFKLWHFWXUH RI WKH 6\VWHP RI %XVLQHVV ,QWHOOLJHQFH
Inmon divides the business intelligence system, as cited by [3], into four-level components and metadata management module. The general architecture of business intelligence system is shown in Fig. 1. The components in Fig. 1 interact with each other to facilitate the fundamental functions of business intelligence: extracting data from the operational system of the company, saving extracted data into data warehouse, and fetching the saved data for various applications of business analysis. Level of operational system; as data sources from the system of business intelligence, business operational system typically uses the system of online transaction processing (OLTP) to support daily business activities.
Level of data acquisition; in this level, a component before the process is divided into three stages: extraction, transformation, and loading (ETL). A company has several systems of OLTP that bring RXW D YHU\ ODUJH QXPEHU RI GDWD 7KH GDWD ¿UVW DUH extracted from OLTP's system of ETL's process and then transformed in accordance with transformation rules. If the data have been transformed, the data, then, are input in data warehouse. ETL is a basic component of business intelligence system because the quality of data from other components depends on the process of ETL. In the planning and developing RI (7/ WKH TXDOLW\ RI GDWD V\VWHP ÀH[LELOLW\ DQG speed process are the main concern. Level of analytical; based on data warehouse, various types of analytical applications have been developed. The system of business intelligence supports 2 basic types in analytical function: reporting and online analytical processing (OLAP). The function of reporting provides managers various types of business reports such as sales report, products report, and human resources report. Report is produced from operating queries into data warehouse. Generally, GDWD ZDUHKRXVH TXHULHV KDYH EHHQ GH¿QHG E\ GDWD warehouse developer. Reports that have been made by business intelligence's system typically have static format and contain exact type of data.
Business intelligence analytical that is very promising is OLAP. According to Codd et al as quoted by [3], OLAP enables managers to go deeper into business data from multi-dimensional analysis through slice, dice, and drilling operations. A dimensional analysis is a perspective through how the data are presented, for instance: type of products, sales location, time, and customers. Compared to report function, OLAP supports data analysis in accordance with needs. OLAP is a model of multidimensional data known as star schema DQG VQRZÀDNH VFKHPD ,Q DGGLWLRQ WR UHSRUWLQJ DQG OLAP, there are a lot of other analytical types that can be made based on the system of data warehouse, such as data mining, executive dashboards, customer relationship management, and business performance management.
Metadata management; metadata is a distinctive data concerning other data, such as data sources, data warehouse saving, business regulations,

DFFHVV DXWKRUL]DWLRQ DQG KRZ GDWD LV H[WUDFWHG DQG
transformed. Metadata is very important in producing accurate and consistent information and system PDLQWHQDQFH 7KH PDQDJHPHQW RI PHWDGDWD LQÀXHQFHV all process from plan, develop, test, deployment and the use of business intelligence system.

%XVLQHVV ,QWHOOLJHQFH LQ %DQNLQJ ,QGXVWU\
The implementation of business intelligence in banking industry is the key success in making the PDLQ EXVLQHVV DFWLYLWLHV HIIHFWLYH DQG HI¿FLHQW ,W KDV WKH DELOLW\ LQ REWDLQLQJ PDQDJLQJ DQG DQDO\]LQJ WKH data of customers, products, services, operational activities, suppliers, and partnerships in a very large numbers. Examples of the implementation in business intelligence in banking industry are customer relationship management, customer credit analysis, risk management, credit card analysis, customer segmentation, etc. [4,5].
The role of business intelligence in business activities can provide more personal services towards customers and radically increases the service quality from the bank. Banking product managers compete in designing products and services that can answer every need in certain segment.
One example of the use of customer credit analysis is the implementation of consumer credit scoring [6]. Consumer credit scoring is the most important activity to evaluate consumer loan application. The credit scoring system is used to model the risk potential from loan application, where WKH V\VWHP KDV EHQH¿WV IRU VZLIWO\ KDQGOLQJ QXPHURXV loan applications without requiring a lot of resources so that it can reduce operational cost and reduce reasoning in decision making effectively. With the competition and growth of consumer credit market, players in banking industry compete each other to develop improved strategy with the help of the implementation of customer credit scoring model.
The purpose of credit scoring is to contribute competence to the part of credit analysis to decide the consumer loan application that has been accepted by the marketing of the bank including a "good credit" where customers included in the FDWHJRU\ KDYH VLJQL¿FDQW SRVVLELOLW\ WR SD\ WKHLU ¿QDQFLDO UHVSRQVLELOLW\ WRZDUGV WKH EDQN RU D ³EDG credit" where costumers included in the category KDYH VLJQL¿FDQW SRVVLELOLW\ WR IXO¿OO WKHLU ¿QDQFLDO responsibility.
Results of study conducted by Ref. [6], researchers compared the performance of credit scoring models using the traditional approach DQG DUWL¿FLDO LQWHOOLJHQFH GLVFULPLQDQW DQDO\VLV ORJLVWLF UHJUHVVLRQ QHXUDO QHWZRUNV FODVVL¿FDWLRQ and regression tree). Experimental research with UHDO GDWD KDV GHPRQVWUDWHG WKDW FODVVL¿FDWLRQ regression tree, and neutral networks overpower the performance of credit scoring model traditionally in terms of prediction accuracy and type II errors. Analysis towards customer data is the key point for the management of the bank to afford maximal revenue. By using pareto concept, designing products and services towards 20% of customers Customer segmentation is an effective marketing strategy, by comprehending the characteristics and needs of every customer segment; then the management may design how to market the price, policy for every product and service in order WR SURYLGH PD[LPXP EHQH¿W > @ 7KH LPSOHPHQWDWLRQ of business intelligence in customer segmentation, the process becomes easier because management can easily identify customers' demographic and geographic. Nevertheless, the management has to spare some time and energy if it wants to know about customers' psychographic and behavior. In addition, the management has to identify necessary attributes, such as ages, occupations, incomes, and genders. The attributes easily and commonly can be measured with RFV (recency, frequency, and value from their transactional behaviors) [8,9].
,W FDQ EH FRQFOXGHG WKDW IXO¿OOLQJ WKH QHHGV RI FXVWRPHUV EHFRPLQJ PRUH FRPSOH[ DQG WKH HI¿FLHQF\ of business processes with automation, operational activities need supports in system information. System information in banking should still be developed in order to meet the needs of customers and follow business innovation. However, it should be integrated into business intelligence system so that the management obtains up-to-date information and insight from historical data.

%XVLQHVV ,QWHOOLJHQFH LQ 5HWDLO ,QGXVWU\
Observing the growth of retail industry in Indonesia, customer relationship management (CRM) is a primary trigger in perceptive businessmen to redesign the focus of their business toward the customers. Retail companies generally have a lot of customers; and the customers commonly have different needs. With the implementation of CRM, then, the management can automate interaction between FXVWRPHUV DQG VDOHV WHDP ,W FDQ DQDO\]H FXVWRPHUV ¶ data obtained from POS transaction, customer service, etc. so that the management may acquire insight towards the customer needs and develop oneto-one relationship with the customers, design and SURPRWLRQDO FDPSDLJQ SURGXFW OD\RXW RSWLPL]DWLRQ Analytical CRM uses business intelligence tools, such as data warehousing, data mining, and OLAP. Some applications of analytical CRM are customer segmentation, campaign/promotion effectiveness analysis, customer lifetime value, customer loyalty analysis, cross selling, product pricing, and target marketing [4]. Some retail companies start to persuade buyers who have not become their member to issue their member card, collaborate with the bank to provide GLVFRXQW HWF 7KH PDQDJHPHQW VWDUWV WR UHDOL]H WKH importance to obtain comprehensive customer data , where the data can give information such as customer characteristic (age, gender, marital status, education, occupation, income per month), customer behavior (customer feedback related to products and services, recommendation from customers related to products and services, substitution products used by customers, customers loyalty toward services from certain brand of a product), and customer expense (purchase price, quantity, frequency of repeat purchases, customers' desire to buy other products and services from a particular manufacturer, etc.) [10]. According to customer's segmentation performed by Ref.
[10], it can be concluded as follows.
&DWHJRU\ LV FXVWRPHUV ZKR ZRUN DW RI¿FH as professional and manager, have a fairly high education, and have house and car. Based on the information obtained, the customers from this category do not have baby, yearly income is not so high but they love shopping especially in fashion. The focus of consumption in this category is high quality FRVPHWLFV RULJLQDO &'V PDJD]LQHV DQG KRXVH ZDUH products in general. Generally, they are not interested in getting food and beverages since their incomes are considerably high. The compatible marketing strategy in this category is to provide higher quality of products and services in order to attract customers because customers from this category are not really interested in discounts.
Category 2 is customers who work in factory as worker and mechanic; and relatively have low income. Some of them have house; shop for food, domestic purposes and toddler' products. Most of them are women who like to go to supermarket to buy more stuff. In general, they are interested in promotional discounts, so that the revenue is not really high.
Category 3 is customers who have low education, low income that almost the same with factory worker. In general, they buy more commodities when there is a substantial discount. Marketing strategy for this category is to provide promotional discount towards the commodities so that it attracts more customers. The implementation of business intelligence within the area of CRM, BI can be applied to supply chain management (SCM). By applying the SCM, the PDQDJHPHQW RI WKH FRPSDQ\ FDQ HI¿FLHQWO\ FRQWURO the inventory and purchasing process to the suppliers. The data taken from purchasing process and supply may contribute various competitive insights for dynamical supply chain. The implementation of data warehouse to SCM can help the management LQ DQDO\]LQJ VXSSOLHUV ¶ SHUIRUPDQFH DQG FRQWUROOLQJ VWRFN OHYHO VDIHW\ VWRFN ORW VL]H DQG OHDG WLPH analysis), product movement, demand forecasting, etc.
The implementation of BI in alternative sales channel can increase the effectivity in managing various types of distribution channels such as Internet, catalog, etc. With the development of today's technology, it enables a customer to interact with the company through various channels in a period. As an example, the development of tablet and smartphone causes the management to expand m-channel to provide additional choices for customers to access the company. The application of BI in alternative sales channel is E-business analysis, web log analysis, referrer analysis, error analysis, keyword analysis, ZHE KRVWLQJ FKDQQHO SUR¿WDELOLW\ DQG SURGXFW FKDQQHO DI¿QLW\

%XVLQHVV ,QWHOOLJHQFH LQ (GXFDWLRQDO ,QGXVWU\
Nowadays the implementation of business intelligence in educational industry is relatively lower than banking industry, health, insurance, etc. [11]. The implementation of business intelligence in educational industry can be done in the process of college admission, teaching management, etc. [12,13]. According to Ref. [12], the competition for college admission is getting tougher every year with most college students receive the application of admissions and more selective in its acceptance. The level of acceptance in well-known university reaches 10%; and the uncertainty causes talented students apply to the school on the next layer. This causes students to insert application to several different schools and every school has different due date. As an effect, students often face with dilemma when they are running out of time to accept the offer from a university that is lower than their priority.
The challenge in the admission process is process of identifying the best applicant including some parameter and when the desired candidate is LGHQWL¿HG WKHQ WKH GHFLVLRQ WR SURSRVH DQ RIIHU DQG the composition of the offer is moderately hard. In addition to the process of students' admission, the application of data mining can be used to support teaching management. Every university manages students' mark in a very large number from different faculties. With the application of data warehouse DQG DQDO\]LQJ WKH GDWD ZLWK YDULRXV GDWD PLQLQJ techniques, the faculty manager can exploit different hidden information and perform forecasting and analysis. So that, faculty managers can use it to improve the quality of teaching and knowledge. variable FODVVL¿FDWLRQ Reference [12] was very sure that data mining and the technique of income management can be used effectively to solve the problem. Data mining is used to expand the model that can predict the quality of applicants using student performance data based on past performance of the students in WKH ¿UVW \HDU LQ WHUPV RI *3$ DQG VRPH LPSRUWDQW parameters gathered from the applicants' data, such as: high school GPA, SAT math score, SAT verbal score, strength of curriculum, adjusted GPA, adjusted test scores, subjective score, and overall assessment score.
This research used neutral networks method since it has better performance compared to decision trees, besides that the capability of neural networks LQ DGDSWLQJ ZLWK VLWXDWLRQDO PRGL¿FDWLRQ PDNHV this method compatible with the context of college admissions. The model of income management is already widely used by companies in the airline LQGXVWU\ DQG WKH KRWHO 7KLV WHFKQLTXH PD[LPL]HV revenue by collecting the best price for each bench/ resource despite of the uncertainty about future demand. This research used dynamical model that is markovian periods since it has capability to handle demands which come randomly. The table of offered price can be used as a reference for the admission staffs to accept or decline application from the students' candidate. Table 2 provides assumption of the total accepted applicants in every week for three categories within 4 weeks, the total of the applicants including the applicants that are accepted and rejected. Table 3 shows offered price for a time period with additional 4000, every 4000 accumulation in a period of approximately 3.3 days.   Table 4 is a database from the students' scores which contain student serial numbers and the result of some several major courses (fundamental of electrical engineering-FEE, electrical machine and drive-EMD, automatic control principle-ACP, automatic control system-ACS, and higher mathematic-HM). To facilitate in data mining usage, the data from Table 4 should be transformed in conditions as IROORZV ,I WKH YDOXH LV EHORZ WKHQ ¿OO LW ZLWK QRW SDVV DQG LI WKH YDOXH LV DERYH WKHQ ¿OO LW ZLWK 1 (pass) by using the decision tree algorithm C4.5. (2) If the value of course FEE (A) passes, then the value of course ACS (C1) generally will pass. The accuracy rate is 86.4% and the covering rate for students is 59.5%. (3) If the value of course FEE (A) fails, and the value of course EMD (B) also fails, then the value of course ACS (C1) typically fails. The accuracy rate is 85.7% and the covering rate for the total of students is 10%. (4) If the value of course FEE (A) fails, but the value of EMD (B) passes, then the value of ACS (C1) can still pass. The accuracy rate is 81.25% and the covering rate for the total of students is 30.5%. Fig. 2: Decision tree to evaluate students' scores [13] With the above approach in evaluating students' scores, then the faculty manager can notice the connection between the students' performance toward the courses of FEE, ACS, and EMD. Based on the result of the evaluation that has been made, EMD course's instructor needs to give more attention to the students who did not pass EMD and ACS.

CONCLUSION
Managements of various industries have to GHYHORS LQIRUPDWLRQ V\VWHP WR IXO¿OO WKH FRPSOH[ stakeholder's needs. In order to use the information obtained from business activity through the information system, it is necessary to develop business intelligence system. In addition, by integrating the obtained insight from historical data, the system of business intelligence may enable the management to anticipate future behavior from the system and allows the modeling of customer behavior. The characteristic of the implementation of business intelligence system is to support the quality and to prompt decision making.