Data Mart Creation and Management

Empowering Users to Create Data Marts

• Users can request the creation of a Data Mart by contacting the Data Warehouse administrator.
• Users can create their own Data Marts using specific software tools.
• Data Mart creation is problem-specific, focused on a single issue or analysis.
• Data Marts require a review to ensure they include relevant variables.

Data Mart Do's and Don'ts: Process

• Consider relevant background and domain knowledge.
• Use checksums to verify data accuracy.
• Cross-check data across tables and source systems.
• Avoid starting analysis or estimations prematurely.
• Define meaningful metadata.
• Transform data into a format suitable for planned analysis methods.
• Explore data with descriptive analysis and graphical representations.
• Perform peer review (independent review).

Data Mart Do's and Don'ts: Handling

• Ensure traceability of every record through ID variables for cross-checking with original data.
• Never drop ID variables, even if not directly useful for analytics.
• Prevent losing relevant cases from datasets.
• Use meaningful units to categorize continuous variables based on statistics or business rules.
• Data Marts represent prepared data from a Data Warehouse for specific uses.

Data Mart Types

• Virtual Data Marts:
  • Create "views" of original data in legacy systems.
  • ETL processes create virtual views rather than real data copies.

Components of a Data Warehouse (DW)

• Data Warehouse (DW) consists of three key components:

  • DBMS: DataBase Management System
  • DB: DataBase
  • DBCS: DataBase Communication System

• DBMS manages metadata relevant to loading, error detection, constraints, and validation.

• DB stores the data.

• DBCS enables data analysis using SQL or other tools.

DataBase Management System (DBMS)

• DBMS focuses on managing the analysis-oriented database.
• Provides functionality for data definition and manipulation.
• Requires different specifications compared to operational systems.

DataBase (DB)

• Data integrity and consistency are evaluated differently in DW than in operational systems, as are security and availability.
• DW data comprises copies of operational datasets with transformations (e.g., cleaning) and additional information extracted.
• DW data often includes more detailed information than operational data.
• DW data integrity, consistency, security, and availability exceed those of operational systems.
• Storage and access optimization in DW is critical for timely responses to complex queries and analyses while maintaining flexibility.

DataBase Communication System (DBCS)

• DBCS plays a crucial role in analysis-oriented systems, enabling effective and efficient utilization of data stored in a DW.
• Detailed metadata/meta-databases are an indispensable component of DW.
• Meta data is essential for effective data analysis.

Data Marts

• Data Marts are distinct from Data Warehouses but are part of analysis-oriented systems.
• Data Marts focus on specific data needs for a particular view or use case.
• Data Marts can be considered subsets of the Data Warehouse, but act as independent entities.
• Data duplication is expected within a Data Mart.

Data, Information, Knowledge, and Wisdom

• Data, information, knowledge, and wisdom are distinct concepts.
• Information serves as a foundation for knowledge needed for specific purposes.
• Knowledge can be seen as something outside of traditional data systems.

Data Capital

• Data capital represents the value associated with data.
• The value of Data Capital is realized by transforming it into information.
• Data Capital is often underutilized.

Analysis-Based Information Systems (ABIS)

• ABIS encompass systems for storing and preparing data for DM/DA.
• Examples include Data Warehouses and Knowledge-Based Information Systems (KBIS).
• ABIS provide tools for loading, transforming, and storing data effectively.
• ABIS include Data Warehouses, Data Marts, and analytical tools, such as Online Analytical Processing (OLAP) and DM tools.

Importance of Information

• Effective information systems enable businesses to adapt to rapid changes in the business environment.
• Information systems provide a competitive advantage.
• Large volumes of captured data can be valuable as "data capital" but may be underutilized or neglected.
• Decision makers often lack access to essential information at critical times or in a form suitable for analysis.
• The 80/20 rule applies in many information systems, indicating that a significant portion of collected data may remain unused in decision making.

Example Questions for Decision Makers

• Which customers should receive specific offers?
• Which customers are at risk of leaving the business?
• What is the cross-selling potential for a new product?