Business Intelligence Chapter 3

Study Notes

Business Intelligence, Analytics, and Data Science: A Managerial Perspective

Chapter 3 focuses on descriptive analytics, business intelligence, and data warehousing.
Business intelligence (BI) used to encompass all data-driven decision support; now it's a part of business analytics, specifically descriptive analytics.
BI is now defined as descriptive analytics.
Descriptive analytics questions what happened and what is happening.
Enables include reporting, dashboards, scorecards, and data warehousing.
Outcomes consist of well defined business problems and opportunities along with business intelligence.
Predictive analytics addresses what will happen and why it will happen.
Prescriptive analytics tackles what should be done and why.
Advanced analytics encompasses optimization, simulation, decision modeling, and expert systems.
A data warehouse is centralized information that enhances informed decisions.
Data flow from transactional systems, relational databases, and other sources is typical.
Data warehouses must have integrated data in standardized format, and be cleansed for enterprise-wide use.
A relational database organizes data into predefined relationships within tables (or relations).
Data in a relational database is organized in columns and rows and makes connections between data structures easy to view and understand.
Data warehouses are subject-oriented, integrated, time-variant, and nonvolatile.
Data in a data warehouse is not updated; changes are stored as new data.
Data warehouses do not go through data quality processes such as cleansing and de-duplication.
Meta-data includes data about the data itself outlining how the data is organized.
Data warehouses allow easy access to end-users through techniques such as web-based, relational/multi-dimensional, and client/server architecture.
Data marts are departmental small-scale data warehouses containing only limited or relevant data (e.g. marketing, operations).
Dependent data marts are subsets directly created from data warehouses.
Independent data marts are alternative lower-cost scaled-down versions of data warehouses.
An independent data mart is designed for strategic business units or departments and its source is not the Enterprise Data Warehouse.
Operational data stores (ODS) are temporary, provisional areas of data warehouses used for short-term decisions, storing recent information in a short-term memory role, different from the long-term memory function of a data warehouse.
ODS consolidates current volatile data from multiple sources.
Operational data marts are created when operational data needs multidimensional analysis.
In multidimensional analysis, data is organized by entities and dimensions.
Time, location and item type are presented in a multidimensional view.
A three-tier data warehouse architecture is comprised of the data acquisition layer, the data warehouse that holds the data, and the client software application tier.
A two-tier architecture combines the data acquisition and data warehouse layer into one.
A single-tier architecture presents the data warehouse as a virtual entity.
ETL (Extract, Transform, Load) is a process comprised of data access, data federation, and change capture to create integrated enterprise data.
ETL processes data from various sources for integration into a data warehouse.
ETL uses several technologies, including Enterprise application integration (EAI), Service-Oriented Architecture (SOA), and near-real-time data delivery.
Data warehousing comprises operational data stores (ODS), Operational data marts, Dependent & Independent Data Marts.
Metadata stores information about the data.
Ten factors influencing architecture selection relate to information interdependence, management information needs, urgency for a data warehouse, and end-user tasks and resource constraints with additional variables such as strategic view of the data warehouse, compatibility with current systems, ability of in-house IT staff, technical issues, and social and political factors.

Data Warehouse Development

Data warehouse development is complex.
Data warehousing affects multiple departments and includes many input and output interfaces.
Direct benefits include end-user analysis capabilities, consolidated corporate data views, enhanced business performance, and simplified data access.
Indirect benefits include enhanced business knowledge, gained competitive advantage, improved customer service and satisfaction, and assistance in reforming business processes.

ETL (Extract, Transform, Load)

Includes extract, transform, and load processes.
ETL tools face challenges such as expensive data transformation tools and a long learning curve.
Important selection criteria include the ability to access various data sources, automatic metadata and open standard conformity, ease of user interaction.