Data Analytics and Business Intelligence Detecting Telecommunications

Question: Discuss about Data Analytics and Business Intelligence for Detecting Telecommunications Fraud in Telstra. Answer: Introduction The research proposes to identify a proficient telecom company from Australia and background of the proposal considers fraudulent activities inside the organization. In order to apply data mining procedures to analyze the organizational vulnerabilities; the proposal required to initiate a suitable methodology (Mwanza and Phiri 2016). The methodology is selected as CRISP-DM for the selected organization Telstra as prominent telecommunication in Australia. In this proposal paper, the primary aim is to identify the objectives of the study, identify the probable outcomes from the study along with mythology justification (NewsComAu 2016). Now, the methodology required to formulate the problem as a data mining scenario, identify the data sources of frauds, and consider the data mining processes. Section 1: Aims, objectives and possible outcomes The aim of the study proposal is to identify the problem area as a potential telecommunication fraud case. The problem area is identified as: Mobile billing scam and fraudulent third-party billing system are constantly fuming the Telstra customers. The additional subscription service was under the advertisement and web page of Telstra, adjourned automatic subscription for several customers. Now, due to this automatic subscription, the complaint dropping from the unhappy customer increased at a rate of 30% than normal complaints (Arora, Gupta and Pahwa 2015). In order to analyze the increasing complaints from customers, their individual speculation of fraud cases and the revenue statistics from subscriptions, the proposal is prepared (NewsComAu 2016). Therefore, the aims of the proposal are included as following: The proposal requires collecting the multidimensional data from Telstra; the data should include the dimensions as subscription time, duration of subscription, location of the customer, and others. The data is essential for conducting the further study over the telecommunication fraud case analysis. The proposal furthermore requires demonstrating the methodology and data mining technique for identifying the fraud activity and behavioral patterns. The fraudulent activity patterns and unusual customer behavioral patterns are a part of the further study after the proposal. As stated aims of the proposal, the objectives are identified to depict the overall proposal importance (NewsComAu 2016). The objectives are listed alongside the aims of the proposal as following: To identify organizational data sources for analysis To state a suitable data mining technique employing the CRISP-DM methodology To demonstrate the fraud case and Telstra background in the study To identify the further data analytics activities under a suitable plan To list all the data analysis activities preparing a timeline of essential procedures Now, the aims and objectives are identified under the Telstra telecommunication fraud contexts, the relevant outcomes of the study are to be depicted under this section (Kabari, Nanwin and Nquoh 2016). The primary part of the study includes collection of contextual datasets from the organization database or trusted sources. The datasets are to be prepared in a suitable format for employing data mining over it to generate analytical reports. The data mining technique is to be identified for employing it under those collected multidimensional data. Finally, the modeling is to be prepared as well upon the data (Wong and Venkatraman 2015). Moreover, the evaluation of the analysis process is to be addressed in the study. Section 2: Background Telstra is identified as largest telecommunication and media organization in Australia. The organization is involved under operation and telecom network building. Besides marketing under mobile, voice calling, data service, payment based television services; the organization is involved in telecom operations as well. As per national scale, the organization Telstra has largest and fastest mobile network in Australia (NewsComAu 2016). The organization is provider of 15.8 million mobile services in Australia along with 2.8 million fixed broadband services for retail sectors. Moreover, it provides 7.7 million fixed voice calling services in parallel. The organization has a global recognition among more than 15 countries including China under the recognition list. Telstra started off a postmaster department collaborating with Australia post earlier, under a government agency (Mwanza and Phiri 2016). The organization was generalized at that time, during the start-up business, now it emerge d as privatized incorporation. It had undergone continuous and relentless transformation for gathering huge sales and marketing strength within a shorter period. Telstra is under regulation of Australian Government being a national broadband network (Arora, Gupta and Pahwa 2015). Not only Telstra is identified as a provider of customer service, however now, the organization is recognized as provider of digital content among the customer as well. It has acquired overall revenue of $6,313 million AUD as estimated in 2013, and still the revenue is increasing in a rapid manner. There are a few sorts of research and exploration works in the area of fraud recognition. Those related studies incorporate misrepresentation location in credit cards transaction, broadcast communications, tax evasion, and interruption discovery (Kabari, Nanwin and Nquoh 2016). Normally the proposed methods use counterfeit consciousness as a rule, utilizing either separately or conjointly arrangements from simulated neural systems, measurable examination, econometrics, master frameworks, fuzzy rationale, hereditary calculations, machine learning, design acknowledgment, representation and others. Wong and Venkatraman (2015) presents wide studies and exchange of exploration with respect to strategies for handling different sorts of cheats. Henecka and Roughan (2015) portrays the instruments accessible for factual extortion identification and the ranges in which misrepresentation discovery advancements are most utilized, calling attention to the basic truth that sometimes one can be sur e, by measurable investigation alone, that an extortion has been executed. Because of this vulnerability, in (Abdallah, Maarof and Zainal 2016), the examination is focused on how databases of client exchanges must be submitted to a few information digging procedures that quest for examples demonstrative of extortion, a procedure which speaks to a test in misrepresentation identification given the need to discover calculations that can figure out how to perceive an awesome assortment of extortion situations and adjust to distinguish and anticipate new situations (Barclay, Dennis and Shepherd 2015). This paper checks these studies and tries to bring a powerful misrepresentation location arrangement in light of choice trees and information mining, with tests on extensive databases logged from media transmission divisions. For Telstra, direct advertising, by post or telecommunications, is expanding as a medium of trade in Australia (Barclay 2015). In the year 2012, telemarketing contained a quarter century penny of the volume of the A$4.5 billion Australian direct advertising industry, and about portion of grown-up Australians have gotten a phone call identifying with telemarketing exercises (Asamoah and Sharda 2015). The more recognizable of these developments are ones which allow the capacity and recovery of phone numbers, programmed rapid dialing, and the transmission of recorded sales. Telemarketing has turned out to be a great deal more productive than direct mailing or way to entryway deals. The phone is being supplemented as a medium of electronic trade by the Internet and by business on-line administrations (Delias, Doumpos and Matsatsinis 2015). All around, at present the Internet comprises of 15,000 PC systems connected to twenty million clients in more than 175 nations, numbers which are gro wing every day. Around a large portion of a million Australians are now associated with the Internet, some through instructive and business endeavors and others secretly at home using modems associated with PCs. Section 3: Data analytics scenario and methodology The Cross Industry Standard Process for Data-Mining is a suitable model for formulating a problem area and statement with data mining process. The data mining is a technique for solving an analytical problem with expert personnel (NewsComAu 2016). The model identifies the different stages of data mining project implementation as depicted underneath. Implementation of CRISP-DM methodology is dependent upon the process flow shown in the figure. The model proceeds with following steps and directions as shown: Business understanding: The step includes understanding all the business rules, regulations, and related objectives for applying this methodology. The primary step is aimed at defining the business aims from the fraud history and database logs (Henecka and Roughan 2015). The proposed target of the study is to detect the fraud cases from the datasets with extracting them for obtaining proper understanding of the situation. Importantly, the understanding is relevant for current telecommunication organization loss due to the fraudulent cases caused to the customers and the company itself (Abdallah, Maarof and Zainal 2016). Moreover, when the implementation of the model is performed, after the implementation, the evaluation process is conducted for related minimization of the issues. The business understanding implementation model assesses the risks under the context for developing a suitable plan (Barclay, Dennis and Shepherd 2015). The planning is required under the further stages of C RISP-DM methodology. Figure 1: CRISP-DM Methodology Process Stages (Source: Delias, Doumpos and Matsatsinis 2015, pp. 370) Data understanding: The second and next stage of the methodology required identification, collection, and description of the relevant database. The initial data sources are to be identified from the current business and organization (Asamoah and Sharda 2015). The data is to be collected under the study; the description should be procured for the data mining purpose. The verification of the data source quality and data format quality is to be provided for reliable analysis process. In this particular analysis stage, fraud case history is depicted and synthesized under the study (Barclay 2015). The required attributes of the formulated data is noted as time of subscription, location of the customer, customer data in details, duration of subscription and more specific fields included. Data preparation: The following step in data mining and process incorporates preparation of the data for importing into a data mining tool or software. In this particular context of the study, the choice of algorithm usage is decision tree utilization (Perez et al. 2015). This particular stage includes finding the calculation fields, external data collection, proper data formatting, cleaning, and classification of the attributes. The attribute classification can be performed under three categories as irrelevant, categorical, and numeric. Modeling: The modeling stage in this methodology employs the different modeling techniques to work over the prepared data from prior stage (Munro and Madan 2016). Intermediate formatted data from data preparation phase, is selected and applied with proper data modeling tools. The tools may be application of neural network, decision tree or other. The particular organization context, is utilized with decision tree for training data, validate data sources and collected data, and testing of fraud cases. Evaluation: The evaluation phase is to conduct a checking process for performing an assessment whether the tool is suitable for continuing the data mining or not (Schmidt, Atzmueller and Hollender 2016). The data is properly utilized under the analysis process, whether the data really portrays the current scenario of fraudulent case in Telstra. Whether, the current data assessment is properly covering the business understanding phase or not, this justification is the key to identify the analysis effectiveness (Jindal et al. 2016). In case, there are more processes that needed to be modeled under the analytics, the process returns to the business understanding phase for reiteration under the process flow. After the implementation of the CRISP-DM method, the deployment is to be conducted accordingly; the deployment phase requires conducting some training sessions with the users of the system (Kavitha and Suriakala 2015). The sessions are a part of making the internal stakeholders understood about the study process. The continuous blank transaction or several transaction logs being produced can forge more fraud cases with a rapid pace. Thus, the project of data mining utilizes more sensitive data that may be true today, may not have any value tomorrow (Niu et al. 2016). Therefore, the collected data are volatile in nature; new types of frauds are expected to happen as well. Section 4: Plan and timetable The study planning includes the decision tree utilization under the current context of Telstra. Decision tree is considered as data representing structure with data mining process. The prior section of CRISP-DM methodology depicts the modeling phase for decision tree based analysis (Abdelhafez and ElDahshan 2015). The main consideration of the decision tree is on the breaking down all the relevant testing process, in form of node and rules. The fundamental technique of Divide to Conquer is employed in the tree structure for analyzing the entire fraud scenario; the problem is segmented under the process. On the other hand, the complexity of any problem can be minimized with decision tree analysis (Steele et al. 2015). The following figure shows the process of a decision tree for analysis. Figure 2: Decision tree analysis structure (Source: Abdelhafez and ElDahshan 2015, pp. 110) The primary segments of the tree are illustrated as: The node: The node comprises of main problem and testing of the scenario. The branch: The branching of the nodes represents the response for individual node. Leaf: The leaf is identified as a class associated node under the structure. Rule: The rule is associated with individual routes from the nodes under classification of the branches. Now, along with application of decision tree analysis tool, the detailed timeline is prepared showing all the relevant activities under the study (Nayak 2015). The activities are scheduled with week basis time span. The timeline is shown as underneath: Project Activities Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week 10 Week 11 Week 12 Week 13 Week 14 Week 15 Week 16 Project Initiation Project Aim Understanding Objectives Identification Project Approach Identification Outline of Methodology Demonstration Data Collection and Source Identification Analysis Tool Identification Project Planning Timeline Specification Activity Scheduling Project Implementation Planning Identification of Milestones Data Analysis Stages Demonstration Implementation of Decision Tree Business Understanding Data Understanding Data Preparation Modeling with Decision Tree Evaluation of The Analysis Data Analysis with Decision Tree Project Completion Report Preparation Results and Discussion Limitation Identification Recommend Best Practices of Analysis Further Scope Identification Table 3: Timeline Scheduling for Research Activities (Source: Created by author) Conclusion In the current scenario of utilizing data under detailed computation modeling, the data mining is an effective tool for analysis. The analysis process in this proposal is initialized for conducting study with analyzing the fraud situations under Telstra. The data should be collected under the foremost stages of data analysis. The entire study is effective with employing suitable analysis and presentation of complex data under certain format. The proposal is suited as proper practice for the further study, as includes a timeline with specifying the relevant activities. The study should be conducted following those activities to gain procedural advantages of staging the research with methodology application. Decision tree analysis tool is selected in order to provide a head start to the further study with related works. The related work is included in form of literature review of other articles to gain detailed knowledge about data mining technique application as how multidimensional d ata can be analyzed. References Abdallah, A., Maarof, M.A. and Zainal, A., 2016. Fraud detection system: A survey.Journal of Network and Computer Applications,68, pp.90-113. Abdelhafez, H.A. and ElDahshan, K.A.R., 2015. Dependence of the data mining techniques on the decision problem nature.International Journal of Data Science,1(2), pp.103-117. Arora, R., Gupta, D. and Pahwa, P., 2015. Fraud Detection Life Cycle Model: A Systematic Fuzzy Approach to Fraud Management.International Journal of Computational Intelligence and Applications,14(02), p.1550010. Asamoah, D. and Sharda, R., 2015. Adapting CRISP-DM Process for Social Network Analytics: Application to Healthcare. Barclay, C., 2015. Data Mining for organizations: Challenges and opportunities for Small Developing States.Knowledge Discovery Process and Methods to Enhance Organizational Performance, p.189. Barclay, C., Dennis, A. and Shepherd, J., 2015. Application of the CRiSP-DM Model in Predicting High School Students examination (CSeC/CXC) Performance.Knowledge Discovery Process and Methods to Enhance Organizational Performance, p.279. Delias, P., Doumpos, M. and Matsatsinis, N., 2015. Business process analytics: a dedicated methodology through a case study.EURO Journal on Decision Processes,3(3-4), pp.357-374. Henecka, W. and Roughan, M., 2015. Privacy-Preserving Fraud Detection Across Multiple Phone Record Databases.IEEE Transactions on Dependable and Secure Computing,12(6), pp.640-651. Jindal, A., Dua, A., Kaur, K., Singh, M., Kumar, N. and Mishra, S., 2016. Decision Tree and SVM-Based Data Analytics for Theft Detection in Smart Grid.IEEE Transactions on Industrial Informatics,12(3), pp.1005-1016. Kabari, L.G., Nanwin, D.N. and Nquoh, E.U., 2016. Telecmmunications Subscription Fraud Detection Using Artificial Nueral Networks.Transactions on Machine Learning and Artificial Intelligence,3(6), p.19. Kavitha, M. and Suriakala, M., 2015. Fraud Detection in Current Scenario, Sophistications and Directions: A Comprehensive Survey.International Journal of Computer Applications,111(5), pp.35-40. Munro, D.L. and Madan, M.S., 2016. Is data mining of manufacturing data beyond first order analysis of value? A case study.Journal of Decision Systems,25(sup1), pp.572-577. Mwanza, M. and Phiri, J., 2016. Fraud Detection on Bulk Tax Data Using Business Intelligence Data Mining Tool: A Case of Zambia Revenue Authority.Intelligence,5(3). Nayak, J.S., 2015. An Efficient Way for Data Mining via Overlay-based Networking for Enhanced Service.International Journal of Computer Applications,123(16). NewsComAu. (2016).Another blow for Telstra customers. [online] Available at: https://www.news.com.au/technology/gadgets/mobile-phones/telstra-customers-fuming-after-being-caught-up-in-mobile-billing-scam/news-story/8a08d59e6ab88cc96a9452fa1efca8f5 [Accessed 27 Aug. 2016]. Niu, K., Gao, Z., Xiao, K., Deng, N. and Jiao, H., 2016, August. A Real-Time Fraud Detection Algorithm Based on Usage Amount Forecast. InInternational Conference of Young Computer Scientists, Engineers and Educators(pp. 87-95). Springer Singapore. Perez, J., Iturbide, E., Olivares, V., Hidalgo, M., Martinez, A. and Almanza, N., 2015. A Data Preparation Methodology in Data Mining Applied to Mortality Population Databases.Journal of medical systems,39(11), pp.1-6. Schmidt, A., Atzmueller, M. and Hollender, M., 2016. Data Preparation for Big Data Analytics: Methods and Experiences. InEnterprise Big Data Engineering, Analytics, and Management(pp. 157-170). IGI Global. Steele, R.M., Agarwal, T., Dirac, L.P. and Qian, J., AMAZON TECHNOLOGIES, INC., 2015.OPTIMIZED DECISION TREE BASED MODELS. U.S. Patent 20,150,379,426. Wong, S. and Venkatraman, S., 2015. Financial Accounting Fraud Detection Using Business Intelligence.Asian Economic and Financial Review,5(11), pp.1187-1207.