This 2008 ״Business Intelligence״ book, written by Ephraim Torben, Ramesh Scharda, Jay Aronson, and David King, is a comprehensive guide to concepts and, more importantly, business intelligence solutions. It incorporates numerous examples and references to additional academic and business sources. Prepared as a textbook, it can also function as a foundation for academic courses, as evidenced by the accompanying references to sources, exercises, and summaries.
The book targets individuals seeking to explore the breadth of business intelligence, whether at an initial academic or practical level, primarily as an introduction. Despite its title, "Business Intelligence – Management Perspective," in my opinion, it is not intended for senior managers. Instead, it caters to those planning to develop business intelligence within their organization. It requires familiarity with various tools, methodologies, and software, enabling them to recognize potential opportunities and possible difficulties.
The book covers the following topics:
Introduction
Data Warehousing
Business Analytics
Data and Text Mining
Business Performance Management
Visualization
Case Studies
In addition to the wealth of information, an appendix includes case studies categorized by organization, sector, business intelligence solution, and implementing software. It is worth examining and exploring organizations similar to those of readers to discover what others have accomplished and promoted.
Happy reading.
Introduction
The need for tools assisting in business intelligence activities has existed for decades. Still, in the last five years, "business intelligence" has gained recognition as a critical term. The emergence of this term is not merely a fashionable imperative but also signifies a change in the level of need leading to business intelligence solutions. The primary needs impacting organizations in recent years where business intelligence can assist include:
Marketing factors: Intensifying competition; emerging global markets; internet e-markets; innovative marketing methods; outsourcing opportunities; and the need for changing real-time transactions.
Customer requirements: Desire for adjustments and product variety; speed of delivery; increased consumer power and reduced loyalty.
Technology: More innovation; numerous products and services; rapid obsolescence; rising information dumping.
Social factors: Government regulations; security and terrorist movements; SOX; increasing social responsibility.
Business intelligence is not a singular solution but rather integrates and supports growth in innovation, planning and strategy, improvement in decision-making processes, and other managerial measures to address these needs.
Defining business intelligence, it serves as an umbrella for infrastructure solutions, tools, databases, applications, and methodologies. Different people attribute different explanations to it, but its primary purpose is to enable interactive access to data, its processing, and the possibility for business managers and experts to conduct meaningful analysis. This analysis allows decision-makers to gain valuable insights and make better decisions.
Business intelligence involves transforming data into information, information into decisions, and decisions into actions.
Key elements, organized according to the book's summary, include:
Data warehouse. Source data.
Business analytics. A collection of tools for processing and analyzing data warehouse data (including data mining).
Business performance management. Performance monitoring and analysis.
User interface.
Key benefits, as per a survey of 510 organizations, are detailed from high to low:
Save time
Single True Version
Improved strategy and planning
Improved tactical decisions
Process optimization
Cost savings
Data Warehousing
A data warehouse is a database created to support decision-making.
Key features:
Built around organizational/business issues.
Integrated (different sources, different databases).
Time-managed; includes historical information.
Not updating; read-only.
The concept on which data warehouses are based:
Sources of information – operational and external.
Data extraction – in personal programs or using commercial software (ETL).
Data loading – also includes transfers and improvement (ETL). Transfer the data to a target environment.
Database – includes summarized and detailed information.
Superdata metadata.
Middleware tools.
Architectural considerations in setting up the data warehouse:
Separation or connection on one data server, application server, and web server hardware.
General or dedicated tabular plinth.
Tabular realization or Star schema realization (facts and dimensions).
Local warehouses (Data Marts) and Central Warehouse (Data Warehouse); Enterprise Data Warehouse that integrates all existing enterprise warehouses. Work without a warehouse and retrieve directly from the operational and other source databases.
The primary considerations influencing the choice of architecture:
The level of independence between business units.
Needs of top management.
The urgency of the need for the data warehouse.
Nature of end-user tasks.
Resource-related constraints.
A strategic look at the data warehouse before its construction.
Compatibility with existing systems.
Early capabilities of the computing team.
Technical issues.
Organizational, political motives.
Data submission is made possible by one of four concepts:
EAI Enterprise Application Interface – Bringing functionality to the data warehouse by source systems.
SOA Service Oriented Architecture – Ditto, but at the discrete level of a particular item.
EII Enterprise Information Interface – fetching real-time data from source databases.
ETL Extract, Transform, and Load is classic software for retrieving, processing, reclamation, transferring, and loading information.
A deeper comparison of the concept of Data Warehouse versus Data Marts:
The father of perception
Data Warehouse – Inmon
Data Marts – Kimball
Overall view
Data Warehouse - Top-down
Data Marts - Bottom-up
Complexity
Data Warehouse – High
Data Marts – Relatively low
Establishment methodology
Data Warehouse – Spiral
Data Marts - Simple. Based on a tabular model
Engaging in physical planning
Data Warehouse – Extensive
Data Mart - Limited
Organize your data
Data Warehouse – By topic
Data Marts – According to processes
User accessibility
Data Warehouse – Limited
Data Marts - High
Main customers
Data Warehouse – Computing People
Data Marts – End Users
Enterprise Location
Data Warehouse – Central, Headquarters
Data Marts – Linked to operational systems
Risk management
There are numerous risks associated with setting up or maintaining a data warehouse that must be recognized for proper prevention and management:
Lack of a specific mission or goal.
Unclear cleanliness level of the source information.
Insufficient qualifications.
Inadequate budget.
Lack of support software.
Incomprehensible source data.
Unmanaged providers.
Departure of key personnel from the project.
Overreliance on new technology.
Geographically dispersed environment.
Weak or insufficiently supportive sponsor.
Low level of computing literacy.
Political problems.
Unrealistic expectations of users/admins.
Architectural and planning risks.
Ambiguity in demarcation and definition of requirements.
Multi-platform challenges.
Loss of a sponsor.
Need for repair in the operational system.
Geographically and culturally distributed staff.
Data warehouse management insights:
The project must align with the organizational strategy and its objectives.
Commitment must be secured early from all levels of management.
Building the data warehouse in a phased manner is recommended.
Transformation capabilities should be integrated into the data warehouse.
Project management should involve collaboration between computer professionals and business experts.
Load only optimized information and data.
Training requirements should not be overlooked.
Maintain political awareness at all times.
For thought: Active Data Warehouse management, also known as Real-time Data Warehouse, is a data warehouse that is updated (almost) in real-time, expanding the field of use and users.
Business Analytics
Business analytics encompasses various applications and techniques designed to collect, store, analyze, and provide access to data, aiding users in making enhanced business and strategic decisions. Many equate business intelligence with this concept.
Business analytics solutions are available at various levels:
Information and knowledge disclosure: through OLAP, ad hoc queries and reports, data mining (including text and web mining), and search engines. The Mining chapter offers a detailed description.
Decision support systems: report generators, team decision support systems, management and organization decision support systems, web analysis tools, overmanagement, statistical analysis, data mining integrated with predictive analytics, applied artificial intelligence, measurement, and performance management. This category also encompasses automated decision systems.
Systems generating visualizations are discussed in a separate chapter. The connection between these systems is not just an additional layer; it also acknowledges them as analytical systems. Visualization enhances understanding and decision-making.
The primary functions of business intelligence tools include reports (fixed and ad hoc), queries, and analyses. Business intelligence system solutions cater to managers and include:
Drill Down
KPIs (Key Performance Indicators) Management
Trend analysis
Unusual reports
CSF Management (Critical Success Factors)
Status reports
Ad-hoc analysis
Examination of different aspects of the data (Slice & Dice)
Principles for selecting OLAP tools (according to Codd):
Multidimensional perception for queries
Accessible online and batch accessibility
Access to remote sources
Support for multiple simultaneous users
Management of sparse matrices
Flexible reporting
Transparency of the database structure to the user
Consistent report performance
General dimensions
Unlimited cross-dimensional operations
Intuitive data processing
No limit to dimension levels and schema.
Business intelligence typically involves working with multidimensional cubes alongside tables, implemented through various methods, from designated databases to tabular databases or multiple combinations.
Advanced business analytics includes data and data mining (described in a separate chapter) and predictive analytical tools. Analytical prediction tools rely on variables tested for individuals or entities to predict future behavior. There are specialized tools for various types of analytical forecasting, tools supporting automated decision-making, and a dedicated subfield for real-time business intelligence.
Data and Text Mining
Data mining is a term that denotes the exploration of information and knowledge within databases. Over recent years, information mining has advanced, relying on data mining algorithms to analyze organizational texts. A novel domain, Web Mining, has emerged within this context, concentrating on uncovering information and knowledge based on activities on a company's website. Data mining is a process that utilizes tools and methodologies from the fields of statistics, mathematics, artificial intelligence, and learning systems to extract and identify information and knowledge from extensive databases. Limited initially to patterns identified in databases, the term now encompasses all facets of automated analytical intelligence.
Characteristics of the domain:
Data is buried within extensive databases.
Visualization tools aid in the analysis and comprehension process.
The miner, typically an end user without programming capabilities, utilizes drill-down and ad hoc queries for debriefing.
Success necessitates surpassing information and knowledge that were previously unknown or unexpected, requiring creative thinking.
Most tools facilitate the transfer of results to spreadsheets (Excel) or other tools for further analysis.
Due to the substantial volume of data, mining processes are generally conducted using parallel work.
Key terms in the field:
Classification is the most common activity in business intelligence, involving analyzing historical information to create a model for predicting future behavior. Conventional tools include neural networks, decision trees, and more.
Clustering – classifying database contents where members of each group share similar qualities. Unlike classification, group demarcation is not predefined and is built during the process.
Association – identifying relationships between items in a typical record (such as items in a shopping cart).
Sequence Discovery – identifying associations over time.
Visualization – gaining insights through illustrating data and information using various graphic means.
Regression is a standard statistical tool that maps information to an expected value.
Forecasting – predicting future values based on templates built on existing knowledge and data.
Data mining finds applications in various sectors: sales, computing, TV/radio, retail, insurance, health, banking, stock exchange trading, airlines, counterterrorism, marketing, manufacturing, government and defense, and police. Main mathematical tools encompass statistical tools, decision trees, CBR (Case-Based Reasoning), neural networks, intelligent agents, genetic algorithms, and more.
Common errors in data mining projects involve selecting an inappropriate problem for the domain, neglecting the sponsor's perspective on data mining, inadequate time for preliminary data preparation, focusing solely on summary data, failure to document and preserve mining processes and interim results, overlooking suspicious data, repeated algorithm runs without readjusting to conditions, unquestioned acceptance of data and business analysis claims, and measuring results differently than the sponsor.
Additional insights (myth-breaking):
Data mining is a multi-step process requiring proactive planning.
Today's technology is suitable for many business environments.
No dedicated database is required as databases progress and evolve.
New tools enable managers from different levels to mine data.
The organization can use data mining if data accurately reflects the business or its customers.
Data mining:
The field deals with mining from sources that are not tabular but textual, primarily assisting in:
Finding information hidden in documents.
Linking documents in different divisions.
Grouping documents according to shared characteristics.
Examples of use: 1) Analysis of resume documents. 2) Detection of recurring malfunctions in malfunction handling systems.
Web mining:
This includes analyzing site content, examining site structure (and relationships between pages and between sites), and analyzing site activity.
Business Performance Management
Business Performance Management (BPM) signifies the evolution beyond Executive Information Systems (EISs) and business intelligence, representing the next generation of decision support systems. In existence for 25 years, this field encompasses technologies, methodologies, metrics, and applications designed to enhance operational and financial performance within the organization. Business Performance Management is a framework for organizing, automating, and analyzing business methodologies, metrics, processes, and systems to promote organizational performance.
BPM delineates a circular process that connects strategy to planning, monitoring, actions, and adjustments. To implement it effectively, it is sometimes necessary to define the strategy, conduct planning, set objectives for various work processes to monitor, and examine how to bring all these elements into action and adjustments, creating a loop back to the strategy.
Studies have demonstrated that world-class leading companies are significantly more efficient than their peers in managing expenses, focusing on operational excellence, implementing complex strategies (both internal and external), and adapting strategy and planning tactics in combination.
Performance measurement insights:
Focus on measuring key factors.
Combine measurements of the past, present, and future.
Integrate indices with the interests of shareholders, employees, partners, suppliers, and other key parties.
Begin indices from a strategic overarching level and reach the low tactical level.
Define objectives of indices based on research and reality, not arbitrarily.
The primary tools for viewing business performance data are the Balanced Scorecard (BSC) or Dashboard. The BSC displays the index and a color/marker indicating its status relative to the desired outcome on a combined page. This makes it easy to identify areas of target achievement, areas with reasonable progress, and areas with difficulties and problems. The Dashboard displays a wealth of information on a single physical page, using integrated panels, with mainly graphic information allowing drill-down/drill-through for more detail.
Various methods for managing and benchmarking exist, with one of the most well-known being the Six Sigma method.
Main applications include budgeting, planning, and forecasting; building and optimizing a profitability model; scorecard applications; financial consolidation; and financial and statutory reporting.
Visualization
The user interface serves the crucial purpose of delivering data and information to users, setting itself apart from conventional operational information systems. It operates as a visualization tool, enabling users to comprehend data better through its presentation and make more informed decisions. In recent years, significant progress has been made in the realm of user interfaces, encompassing two main types:
Excel-like spreadsheets: Excel proves to be a simple and convenient tool for many users.
Dashboards displayed in portals (panel-based).
A recent addition to this landscape is another type:
This category is based on geographic data, utilizing GIS systems or more straightforward means.
Within this expansive framework, various graphic means exist to illustrate the essence of data and significantly enhance practical understanding. Illustration caters to all users, with its primary applications favoring:
Scorecards and Dashboards for administrators.
Tools tailored for finance professionals.
A decision-making tool for managing risks, applicable to managers and anyone involved in the field.
Case Studies
Toyota Motor Sales USA
Sector: Automotive industry
Business Intelligence Solution:
Dashboards
Data Warehouse
Software:
Oracle Database
Hyperion
Details:
Supply Chain
Sales
Longs Drug Stores
Sector: Retail
Business Intelligence Solution:
Automated Decision Support
SAS
Details:
Automatic pricing
State of Texas
Sector: Government
Business Intelligence Solution:
Data mining
Software:
SPSS
Details:
Improving tax collection
France Telecom
Sector: Telecom
Business Intelligence Solution:
BICC
Software:
Business Objects
Details:
Improving BI utilization and solution uniformity
BNSF Trains
Sector: Transport
Business Intelligence Solution:
Dashboard
Software:
Teradata
Details:
Supply Chain Forecasts
Continental Airlines
Sector: Transport
Business Intelligence Solution:
Real-time DW
Software:
Teradata
Details:
Accounting Management
CRM
Team operations and payroll
Security & Scams
Flight operation
First America's Corporation
Sector: Banking
Business Intelligence Solution:
Data warehouse
Software:
Vision
Details:
Customer-based management
Bank of America
Sector: Banking
Business Intelligence Solution:
Integrated EDW
Software:
Teradata
Details:
Customer-based management
Streamline processes
Hokuriku Coca-Cola Bottling Company
Sector: Retail
Business Intelligence Solution:
Data warehouse
Analytics
Software:
Teradata
Details:
Sales processes, inventory, and exceptions
HP
Sector: Hi-Tech (hardware)
Business Intelligence Solution:
Integrated Data Warehouse
Software:
Not specified
Details:
Unify data marts to understand the big picture
Egg Plc
Sector: Banking (online)
Business Intelligence Solution:
Near real-time data access
Software:
Sun
Oracle
SAS
Details:
Customer-based management
Overstock.com
Sector: Retail (online)
Business Intelligence Solution:
Real-time Data Warehouse
Analytics
Software:
Teradata
Synopsis
Details:
Operational reports
Manage marketing campaigns
Lexmark
Sector: Hi-Tech (hardware)
Business Intelligence Solution:
Business Analytics
Software:
Microstrategy
Details:
Sales Management
Ben & Jerry's
Sector: Food
Business Intelligence Solution:
Business Analytics
Software:
Oracle DW
Details:
Consumption management
TCF Financial Corp.
Sector: Banking
Business Intelligence Solution:
Reports
Data mining
OLAP
Software:
Computer Science
Details:
Cross-selling
Inrix
Sector: High Tech
Business Intelligence Solution:
Business Analytics
GIS
Software:
Dedicated software and algorithms
Details:
Transportation load forecasting
Merrill Lynch
Sector: Financial Services
Business Intelligence Solution:
Data Visualization
Software:
Not specified
Details:
Graphs for agents for personal use or sharing with customers
(General)
Sector: Online retail
Business Intelligence Solution:
Web analytics
Software:
Not specified
Details:
Analysis of consumer behavior for advertising purposes and sales growth
Arkansas State Government
Sector: Government
Business Intelligence Solution:
Geographic information
Software:
Oracle
Details:
Making effective planning decisions
Assistance to geographic information in general
Highmark
Sector: Medical information
Business Intelligence Solution:
Data mining
Software:
Not specified
Details:
Forecasting at-risk populations
Adequate medical care and more
First Health Group
Sector: Medical Services
Business Intelligence Solution:
Data mining
Software:
Not specified
Details:
Customer Service
National Safety Highway Traffic Administration
Sector: Government, Transport
Business Intelligence Solution:
Data mining
Software:
Not specified
Details:
Analysis of accident factors and prone drivers
Mayo Clinic
Sector: Medicine
Business Intelligence Solution:
Data mining
Software:
IBM
Details:
Predicting drug responses
Tailoring optimal care
Department of Homeland Security
Sector: Government, Security
Business Intelligence Solution:
Data mining
Text mining
Web analytics
Software:
Not specified
Details:
Identification of Budgeting for Terrorist Activity
Identify relationships between groups.
Aer Lingus
Sector: Aviation
Business Intelligence Solution:
Text mining
Software:
Megaputer
Details:
Analysis of accident and malfunction reports
HP
Sector: Hi-Tech (hardware)
Business Intelligence Solution:
Text mining
Software:
Temtec
SAS
Details:
Customer Relations
Improve sales
Cisco
Sector: Hi-Tech (hardware)
Business Intelligence Solution:
Business Performance Management
Software:
Not specified
Details:
Manage strategy-dependent metrics
Euro Disney
Sector: Entertainment
Business Intelligence Solution:
Business Performance Management
Software:
Not specified
Details:
Meeting goals
International Truck and Engine Corporation
Sector: Industrial – Automotive
Business Intelligence Solution:
KPI Portal
Software:
Hyperion
Details:
Manage performance metrics
Emergency Medical Associates
Sector: Medicine
Business Intelligence Solution:
Dashboard
Software:
Business Objects
Details:
Performance indicators for managing emergency medicine clinics and medical staff
City of Albuquerque
Sector: Government
Business Intelligence Solution:
Business Activity Monitoring
Software:
Not specified
Details:
Management of operational metrics: alerts, performance anomalies, operational events
Western Digital
Sector: Hi-Tech (hardware)
Business Intelligence Solution:
Dashboard
Software:
Not specified
Details:
Manage performance metrics
Comments