The Power of Business Intelligence:: 11 Works Cited
Length: 3296 words (9.4 double-spaced pages)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
As the business environment changes and becomes more complicated, enterprises are under huge stress to respond and be innovative to such conditions. Enterprises decision making needs to quick and strategic and so making such decisions can be very complex. What this report of Business intelligence (BI) will describe is the tools available to manager to support such decisions, the possible benefits and the limitations of BI. (Turban et al., 2011)
Turban et al., (2011, p. 28) describes business intelligence (BI) as ‘an umbrella term that combines architectures, tools, databases, analytical tools, applications, and methodologies. It is a content-free expression, so it means different things to different people’.
The key goal of BI is to allow for interactive access to data which can be in real-time, easy manipulation of the data, and the ability for management to be able to do suitable analysis of the data. Managers are then able to make more accurate and better decisions through BI by looking at old and new data. (Turban et al., 2011)
Ultimately, Business intelligence has the ability to simplify how managers access and analyse data which makes understanding, collaborating, and acting on information at any point much simpler for decision makers. (Microsoft, 2014)
Turban et al., (2011, p. 30) states that ‘A Business system has four major components’.
1. Data Warehouse
2. Business Analytics
3. Business Performance Management
4. User Interface
Figure 1: Business Intelligence Overview (Turban et al., 2011, p.29)
Turban et al., (2011, p. 52) describes a data warehouse as being ‘a pool of data produced to support decision making; it is also a repository of current and historical data of potential interest to managers throughout the organisation’. Turban et al., (2011, p. 52) went on to state that ‘data are usually structured to be available in a form ready for analytical processing activities (i.e. data mining, querying, reporting and other support applications)’.
In order to conduct businesses processes in the correct manner it is important to use various tools such as BI tools and a combination of real-time data warehousing (RDW) with decision support system (DSS). By having appropriate, dependable information about current trends, changes etc… decision makers are able to make better choices. Data warehousing has the ability to give managers crucial operational data in a reliable, timely and dependable manner. (Turban et al., 2011)
Characteristics of Data Warehousing:
• Subject orientated: e.g. sales, which will only have significant information to support decision making
• Time Variant (time series): Data warehouses must take into consideration time because it is one of the most important dimensions
• Web based
• Real time: Ability to access and examine data in real-time
• Metadata: How the data is structured (Turban et al., 2011)
A data mart consists normally of just one area e.g. marketing. It is a division of a data warehouse and is generally smaller. A data mart can be dependent i.e. a division which is generated directly from a data warehouse and supports the idea of a ‘single enterprise-wide data model’, but firstly the data warehouse must be built or independent i.e. cheaper and smaller type of data warehouse. It is configured to look at a strategic business unit. (Turban et al., 2011)
Operational Data Stores (ODS) & Enterprise Data Warehouses (EDWs)
An ODS is a database which is used as a provisional stage for a data warehouse. The contents of an ODS are constantly updated right the way through the development of business operations, unlike the unchanging subjects of a data warehouse. You will find that an ODS is only generally used for short-term decision making which is concerned with ‘mission-critical’ functions. Alternatively an EDW is a much larger data warehouse which is used right the way across the organisation for decision support and providing effective BI in comparison to ODS’s. (Turban et al., 2011)
Data Warehousing Framework
Figure 2: Data Warehousing Framework (Logic, Fulcrom, 2009)
Data Warehousing Architectures
Three-tier and two-tier architectures are mostly used for data warehousing, but there can be just one-tier architectures. (Turban et al., 2011)
Hoffer et al., (2007) divided the data warehouse into three parts:
1. Data acquisition software (back-end)
2. The data warehouse that contains the data & software
3. Client (front-end) software that allows
Turban et al., (2011, p. 58) shows that in a three-tier architecture, ‘operational systems contain the data and the software for data acquisition in one tier (i.e., the server), the data warehouse is another tier, and the third tier includes the DSS/BI/BA engine (i.e., the application server) and the client’.
Figure 3: Three-tier Data Architecture (GNU, 2009)
The DSS engine in two-tier architecture operates on an identical hardware program as the data warehouse. (Turban et al., 2011)
Figure 4: Two-tier Data Architecture (GNU, 2009)
Alternative Data Warehousing Architectures
The following is a list of alternative to simple architectural design types which are neither EDW nor DW as identified by Turban et al., (2011, p.61)
1. ‘Independent Data Marts Architecture
2. Data Mart Bus Architecture with Linked Dimensional Data Marts
3. Hub-and-Spoke Architecture
4. Centralised Data Warehouse Architecture
5. Federal Architecture’
Data Integration and the Extraction, Transformation, and Load (ETL) Process
Data integration contains three main procedures; data access, data federation, and change capture. When these procedures are properly put in place, they allow data to be retrieved and made available to a range of ETL and analysis tools. (Turban et al., 2011)
One of the main purposes of a data warehouse is to incorporate data from numerous systems. Some of the integration technologies which allow for data and metadata incorporation as described by Turban et al., (2011, p.67) include:
• Enterprise application integration (EAI)
• Service-orientated architecture (SOA)
• Enterprise information integration (EII)
• Extraction, transformation, and load (ETL)
Extraction, Transformation, and Load (ETL)
ETL is at the fore front of the technical side of data warehousing. ETL technologies are vital in the data warehousing process and are an essential element in any ‘data-centric project’. Data transformation happens by using such tools as lookup tables or by linking data with other data. The three functions, extract, transformation, and load are combined into one tool to extract data out of databases and puts them into another united database or a data warehouse. (Turban et al., 2011)
Figure 5: ETL Process (IMC, 2011)
Data Warehouse Development
A data warehousing project is a very complex and large task for any corporation because it is made up of many departments and influences these departments. It is also more difficult than a basic, ‘mainframe selection and implementation project’ because it has a lot of input and output interfaces and it may be included in a CRM business strategy. (Turban et al., 2011)
OLAP versus OLTP
Turban et al., (2011, p. 77) has identified OLTP as the term ‘used for transaction system that is primarily responsible for capturing and storing data related to day-to-day business functions such as ERP, CRM, SCM, POS, and so on’. Turban et al., (2011, p. 77) continues on stating that OLTP systems address ‘a critical business need, automating daily business transactions and running real-time reports and routine analysis’. The downside to OLTP systems are they are not intended for ‘ad hoc analysis' and multiple data items relating to difficult queries.
OLAP, according to Turban et al., (2011, p. 77) are ‘designed to address this need by providing ad hoc analysis of organisational data much more effectively and efficiently’.
Various OLAP operations which are most common include; slice and dice, drill down, roll up, and pivot. (Turban et al., 2011)
Data Warehousing Implementation Issues
Turban et al., (2011, p. 80) as citied in Reeves (2009) and Solomon (2005) have given advice on guidelines regarding questions that must be asked, ‘risks that should be weighted’, and process that can be done to achieve effective data warehouse implementation . The following shows 11 major tasks that could be done in parallel which they have identified:
1. ‘Establishment of service-level agreements and data-refresh requirements
2. Identification of data sources and their governance policies
3. Data quality planning
4. Data model design
5. ETL tool selection
6. Relational database software and platform selection
7. Data transport
8. Data conversion
9. Reconciliation process
10. Purge and archive planning
11. End-user support’.
Real-Time Data Warehousing
In business you often need to make quick and reliable decisions across the organisation but this requires more than just a traditional data mart or data warehouse. Real-time data warehousing (RDW) allows you to load and view data through the data warehouse as they happen. (Turban et al., 2011)
RDW does have its own issues when creating one and these include:
• Not all data should be updated continuously
• Mismatch of reports generated minutes apart
• May be cost prohibitive
• May also be infeasible
Figure 6: Real-time Data Warehousing (Raman, 2012)
Business Management Performance
Business Management Performance is a real-time system that warns management about impeding threats, opportunities and problems. Turban et al., (2011)
Figure 7: Business Management Performance Systems (Watters, 2014)
Figure 8: Business Management Performance Systems (ii) (Watters, 2014)
Data Mining for Business Intelligence
Data mining according to Turban et al., (2011, p. 151) ‘is a way to develop BI from data that an organisation collects, organises, and stores’. Turban et al., (2011) goes on to state that organisations use data mining as a tool to improve their knowledge of their consumers, their own internal procedures and work out difficult organisational problems.
Characteristics and Objectives of Data Mining
Turban et al., (2011) identified the characteristics and objectives of Data Mining as:
• Often data are contained inside very big databases which occasionally comprises of data from several years
• The data mining environment is usually a client/server architecture or a Web-based information systems architecture. Usually data mining is a ‘client/server architecture’ or a ‘Web-based information systems architecture’
• High-tech new tools which include sophisticated ‘visualisation tools’ and aid in the removal of ‘information ore’ hidden in corporate files or public archives. Furthermore, pioneering data miners are looking into the practicality of soft data e.g. Text files on the internet.
• The end user who is generally the miner has the ability to ask ad hoc questions and retrieve answers rapidly with no previous programing skills.
• End users are able to ‘strike it rich’ and achieve unexpected results if they think in a creative manner throughout the data mining process.
• Data mining tools are easily linked in with various software development tools and spreadsheets (Turban et al., 2011)
Figure 9: Data Mining Elements (SEO & PPC Management Solution, 2014)
How Data Mining Works
By using available and appropriate data, data mining is able to construct models to recognise patterns amongst the elements in the dataset. These patterns are either explanatory or predictive. (Turban et al., 2011)
Turban et al., (2011) identified four major types of patterns which data mining seeks to identify:
1. ‘Associations (finding the frequently ‘co-occurring of things’ )
2. Clusters (identify natural groupings of things based on their known characteristics)
4. Sequential relationships (discover time-ordered events)’.
Data Mining Applications
When it comes to data mining applications the objective of these applications is to answer a problem or to look at new business opportunities in order to build a competitive advantage which can be sustained. (Turban et al., 2011)
Customer relationship management (CRM)
Turban et al., (2011, p. 165) describes the objective of CRM ‘is to create one-on-one relationships with customers by developing an intimate understanding of their needs and wants’. CRM is the newest development to come from traditional marketing.
According to Turban et al., (2011, p. 166) data mining has the ability to help banking in the following ways:
1. ‘Automating the loan application process
2. Detecting fraudulent credit card and online banking transactions
3. Identifying ways to maximise customer value by selling them products and services that they are most likely to buy
4. Optimising the cash return’.
Manufacturing and Production
Data mining can be used by manufacturers to forecast any machine failures before they actually happen, detect any irregularities and cohesions in production systems. (Turban et al., 2011)
Data Mining Process
A general process is normally followed when carrying out data mining projects. One process which is considered the most popular is the ‘Cross-Industry Process for Data Mining (CRISP-DM)’. Turban et al., (2011, p. 169) describes the CRISP-DM process as an arrangement of six steps which follows a sequence starting from ‘good understanding of the business and the need for data mining projects and ends with the ‘deployment of the solution that satisfies the specific business need’.
Figure 10: Data Mining Process (Kyran, 2013)
Other data mining standardised processes include:
• Sample, explore, modify, model, and assess (SEMMA)
• Knowledge discovery in databases (KDD)
Data Mining Methods
When you perform data mining, there are a whole host of methods you can use which include; classification, regression, clustering, and association.
Classification is generally the most used method for real-life problems. By analysing historical data, classification is able to learn patterns so that it can place ‘new instances’ into their correct place. Some classification tasks include; credit approval and target marketing. (Turban et al., 2011)
Turban et al., (2011, p. 178) has identifies factors which are considered in assessing the model:
• ‘Predictive accuracy
Several techniques are used for the classification method that have been identified by Turban et al., (2011, p. 181) which include:
• ‘Decision tree analysis
• Statistical analysis
• Neural networks
• Case-based reasoning’.
Cluster Analysis for Data Mining
Turban et al., (2011, p. 184) has described cluster analysis as being ‘an essential data mining method for classifying items, events, or concepts into common groupings called clusters.’ It can be used for such things as market segmentation of customers and fraud detection. The objective of cluster analysis according to Turban et al., (2011, p. 185) is to ‘sort cases (e.g. people, things, events) into groups, or clusters, so that the degree of association is strong among members of the same cluster and weak among members of different clusters’.
Turban et al., (2011, p. 185) has provided the following information about possible uses for cluster analysis results:
• ‘Identifying a classification scheme
• Suggesting statistical models to describe populations
• Indicating rules for assigning new cases to classes for identification, targeting, and diagnostic purposes
• Providing measures of definition, size, and change in what were previously broad concepts
• Finding typical cases to label and represent classes
• Decreasing the size and complexity of the problem space for other data mining methods
• Identify outliers in a specific domain’.
Artificial Neural Networks for Data Mining
Neural networks have emerged as advanced data mining tools in cases where other techniques may not produce satisfactory solutions. Neural networks have been shown to be very promising computational systems in many forecasting and business classification applications due to their ability to ‘learn’ from the data, their nonparametric nature, and their ability to generalise. Neural network computing is a key component of any data mining tool kit.
Data Mining Myths and Blunders
Turban et al., (2011, p. 199) has identified data mining as being ‘a powerful analytical tool that enables business executives to advance from describing the nature of the past to predicting the future’. Turban et al., (2011, p.199) goes on to state that data mining helps marketers in the process of determining customer behaviours and results from data mining can help organisations to ‘increase revenue, reduce expenses, identify fraud, and locate business opportunities, offering a whole new realm of competitive advantage’.
According to Turban et al., (2011, p. 199) there are many myths associated with data mining, including:
• ‘Data mining provides instant, crystal ball like predictions
• Data mining is not yet viable for business applications
• Data mining requires a separate, dedicated database
• Only those with advanced degrees can do data mining
• Data mining is only for large firms that have lots of customer data’.
Text and Web Mining
According to Turban et al., (2011, p. 212), ‘text mining is the semiautomated process of extracting patterns (useful information and knowledge) from large amounts of unstructured data sources’. Turban et al., (2011) goes on to state that text mining is similar to data mining in the way that it has the same aim and processes as data mining but the key difference being with text mining, the process is made up of various unstructured data files e.g. word documents and pdf files.
Turban et al., (2011, p. 213) identifies the following as being among the most popular application areas of text mining:
• ‘Information extraction
• Topic tracking
• Concept linking
• Question answering’.
Natural Language Processing (NLP)
NLP is an essential element of text mining. It looks at how the natural human language is understood and the problems involved in trying to convert human language into an easier formal language for computer programs to understand and manipulate. The objective of NLP is to be able to truly understand and be able to process the natural human language which allows for both grammatical and semantic constraints along with the actual context. (Turban et al., 2011)
Text Mining Applications
With the increase in the quantity of unstructured data gathered by organisations, the ‘value propositions’ and popularity in using text mining tools also increases. Turban et al., (2011, p. 220) has identified the following as some examples of application categories of text mining:
• ‘Marketing Applications
• Security Applications
• Biomedical Applications
• Academic Applications’.
Text Mining Process
Figure 11: Text Mining Process (Deshpande, 2013)
Turban et al., (2011, p. 237 ) identifies web mining ‘as the process of discovering intrinsic relationships from Web data, which are expressed in the form of textual, linkage, or usage information’.
Figure 12: Web Mining Process (Dürsteler, 2014)
BI Implementation Factors
A large number of factors may influence BI implementation. These factors are technological, administrative, behavioural, and so on. The following are the major factors that affect the decision-making process of BI implementation:
1. Reporting and Analysis Tools
3. Extraction, Transformation, and Load (ETL) Tools
4. Costs Involved
The following are the critical success factors for a BI implementation
• Business driven methodology
• Clear vision and planning
• Committed management support and sponsorship
• Data Management and quality issues
• Mapping the solutions to the user requirements
• Performance considerations of the BI system
• Robust and extensible framework
When it comes to BI there are many benefits associated with it and all the tools, applications, methods etc… The key benefit of BI to an organisation is that it facilitates precise information when necessary.
Other Benefits identified by Turban et al., (2011, p.255) include:
• ‘Time savings and operational efficiencies
• Lower cost of operations
• Improved customer service and satisfaction
• Improved operational and strategic decision making
• Improved employee communications and satisfaction
• Improved knowledge sharing’.
The Limitations of Traditional BI
Although there are clear benefits to BI, there are also limitations. Initially there may be a negative ROI for reasons such as; costly fees to implement BI, licence fees, not being able to provide for its goals and objectives that were first set out and maintenance/hidden costs throughout the life of any BI project. Also, in-house BI vendors of traditional BI have rarely given executives the ability to tackle changing issues in real-time when it comes to unified reporting and analysis solutions. (Turban et al., 2011)
Some data mining may also provide for legal/privacy issues because of individuals privacy being invaded. Very often it is hard to justify the intangible benefits of BI and very often there can be ethical issues related to BI implementation because of the privacy issues and accountability of vendors. (Turban et al., 2011)
Deshpande, B. (2013, June 12th). Sima Fore. Retrieved April 30th, 2014, from http://www.simafore.com/blog/bid/116340/Keyword-clustering-using-web-mining-and-text-mining-with-RapidMiner
Dürsteler, J. C. (2014). InfoVis. Retrieved April 30th, 2014, from http://www.infovis.net/printMag.php?num=172&lang=2
GNU. (2009, November 24th). Argo Wiki. Retrieved April 27th, 2014, from http://www.argowiki.com/index.php?title=Data_Warehouse_Management
IMC. (2011). IMC. Retrieved April 28th, 2014, from http://www.imc.com/services/enterprise-data-warehousing/etl-process-management
Kyran. (2013, June 25th). Statistica News and Blog. Retrieved April 30th, 2014, from http://www.statsoft.com/support/blog/entryid/540/crisp-data-mining-session-2
Logic, Fulcrom. (2009). Fulcrom Logic. Retrieved April 27th, 2014, from http://www.fulcrumlogic.com/data_warehousing.shtml
Microsoft. (2014). Microsoft Share Point. Retrieved April 26th, 2014, from http://technet.microsoft.com/en-us/library/cc811595(v=office.12).aspx
Raman, S. (2012, November 29th). BMC Software. Retrieved April 29th, 2014, from https://communities.bmc.com/docs/DOC-10054
SEO & PPC Management Solution. (2014). The Commerce Solution. Retrieved April 29th, 2014, from http://www.theecommercesolution.com/usefull_links/KBS_Data_Mining.php
Turban, E., Sharda, R., Delen, D., & King, D. (2011). Business Intelligence, A Managerial Approach, Second Editiion. Pearson Education, Inc.
Watters, Brian. (2014) Business Management Performance Systems Notes.