DATA WAREHOUSE TERMINOLOGY[1]
Bruce W. Johnson, M.S.
Ad Hoc Query:
A database search that is designed to extract specific information from a database. It is ad hoc if it is designed at the point of execution as opposed to being a “canned” report. Most ad hoc query software uses the structured query language (SQL).
Aggregation:
The process of summarizing or combining data.
Catalog:
A component of a data dictionary that describes and organizes the various aspects of a database such as its folders, dimensions, measures, prompts, functions, queries and other database objects. It is used to create queries, reports, analyses and cubes.
Cross Tab:
A type of multi-dimensional report that displays values or measures in cells created by the intersection of two or more dimensions in a table format.
Dashboard:
A data visualization method and workflow management tool that brings together useful information on a series of screens and/or web pages. Some of the information that may be contained on a dashboard includes reports, web links, calendar, news, tasks, e-mail, etc. When incorporated into a DSS or EIS key performance indicators may be represented as graphics that are linked to various hyperlinks, graphs, tables and other reports. The dashboard draws its information from multiple sources applications, office products, databases, Internet, etc.
Cube:
A multi-dimensional matrix of data that has multiple dimensions (independent variables) and measures (dependent variables) that are created by an Online Analytical Processing System (OLAP). Each dimension may be organized into a hierarchy with multiple levels. The intersection of two or more dimensional categories is referred to as a cell.
Data-based Knowledge:
Factual information used in the decision making process that is derived from data marts or warehouses using business intelligence tools. Data warehousing organizes information into a format so that it represents an organizations knowledge with respect to a particular subject area, e.g. finance or clinical outcomes.
Data Cleansing:
The process of cleaning or removing errors, redundancies and inconsistencies in the data that is being imported into a data mart or data warehouse. It is part of the quality assurance process.
Data Mart:
A database that is similar in structure to a data warehouse, but is typically smaller and is focused on a more limited area. Multiple, integrated data marts are sometimes referred to as an Integrated Data Warehouse. Data marts may be used in place of a larger data warehouse or in conjunction with it. They are typically less expensive to develop and faster to deploy and are therefore becoming more popular with smaller organizations.
Data Migration:
The transfer of data from one platform to another. This may include conversion from one language, file structure and/or operating environment to another.
Data Mining:
The process of researching data marts and data warehouses to detect specific patterns in the data sets. Data mining may be performed on databases and multi-dimensional data cubes with ad hoc query tools and OLAP software. The queries and reports are typically designed to answer specific questions to uncover trends or hidden relationships in the data.
Data Scrubbing:
See Data Cleansing
Data Transformation:
The modification of transaction data extracted from one or more data sources before it is loaded into the data mart or warehouse. The modifications may include data cleansing, translation of data into a common format so that is can be aggregated and compared, summarizing the data, etc.
Data Warehouse:
An integrated, non-volatile database of historical information that is designed around specific content areas and is used to answer questions regarding an organizations operations and environment.
Database Management System:
The software that is used to create data warehouses and data marts. For the purposes of data warehousing, they typically include relational database management systems and multi-dimensional database management systems. Both types of database management systems create the database structures, store and retrieve the data and include various administrative functions.
Decision Support System (DSS):
A set of queries, reports, rule-based analyses, tables and charts that are designed to aid management with their decision-making responsibilities. These functions are typically “wrapped around” a data mart or data warehouse. The DSS tends to employ more detailed level data than an EIS.
Dimension:
A variable, perspective or general category of information that is used to organize and analyze information in a multi-dimensional data cube.
Drill Down:
The ability of a data-mining tool to move down into increasing levels of detail in a data mart, data warehouse or multi-dimensional data cube.
Drill Up:
The ability of a data-mining tool to move back up into higher levels of data in a data mart, data warehouse or multi-dimensional data cube.
Executive Information Management System (EIS):
A type of decision support system designed for executive management that reports summary level information as opposed to greater detail derived in a decision support system.
Extraction, Transformation and Loading (ETL) Tool:
Software that is used to extract data from a data source like a operational system or data warehouse, modify the data and then load it into a data mart, data warehouse or multi-dimensional data cube.
Granularity:
The level of detail in a data store or report.
Hierarchy:
The organization of data, e.g. a dimension, into a outline or logical tree structure. The strata of a hierarchy are referred to as levels. The individual elements within a level are referred to as categories. The next lower level in a hierarchy is the child; the next higher level containing the children is their parent.
Legacy System:
Older systems developed on platforms that tend to be one or more generations behind the current state-of-the-art applications. Data marts and warehouses were developed in large part due to the difficulty in extracting data from these system and the inconsistencies and incompatibilities among them.
Level:
A tier or strata in a dimensional hierarchy. Each lower level represents an increasing degree of detail. Levels in a location dimension might include country, region, state, county, city, zip code, etc.
Measure:
A quantifiable variable or value stored in a multi-dimensional OLAP cube. It is a value in the cell at the intersection of two or more dimensions.
Member:
One of the data points for a level of a dimension.
Meta Data:
Information in a data mart or warehouse that describes the tables, fields, data types, attributes and other objects in the data warehouse and how they map to their data sources. Meta data is contained in database catalogs and data dictionaries.
Multi-Dimensional Online Processing (MOLAP):
Software that creates and analyzes multi-dimensional cubes to store its information.
Non-Volatile Data:
Data that is static or that does not change. In transaction processing systems the data is updated on a continual regular basis. In a data warehouse the database is added to or appended, but the existing data seldom changes.
Normalization:
The process of eliminating duplicate information in a database by creating a separate table that stores the redundant information. For example, it would be highly inefficient to re-enter the address of an insurance company with every claim. Instead, the database uses a key field to link the claims table to the address table. Operational or transaction processing systems are typically “normalized”. On the other hand, some data warehouses find it advantageous to de-normalize the data allowing for some degree of redundancy.
Online Analytical Processing (OLAP):
The process employed by multi-dimensional analysis software to analyze the data resident in data cubes. There are different types of OLAP systems named for the type of database employed to create them and the data structures produced.
Open Database Connectivity (ODBC):
A database standard developed by Microsoft and the SQL Access Group Consortium that defines the “rules” for accessing or retrieving data from a database.
Relational Database Management System:
Database management systems that have the ability to link tables of data through a common or key field. Most databases today use relational technologies and support a standard programming language called Structured Query Language (SQL).
Relational Online Analytical Processing (ROLAP):
OLAP software that employs a relational strategy to organize and store the data in its database.
Replication:
The process of copying data from one database table to another.
Scalable:
The attribute or capability of a database to significantly expand the number of records that it can manage. It also refers to hardware systems and their ability to be expanded or upgraded to increase their processing speed and handle larger volumes of data.
Structured Query Language (SQL):
A standard programming language used by contemporary relational database management systems.
Synchronization:
The process by which the data in two or more separate database are synchronized so that the records contain the same information. If the fields and records are updated in one database the same fields and records are updated in the other.
About the Author:
Bruce W. Johnson, MS, PMP is the CEO of Johnson Consulting Services, Inc. He is an information management consultant who specializes in working with social service, healthcare and government agencies. He can be reached at (800) 988-0934 or by e-mail at .
Page 6
[1] Adapted by permission from Behavioral Health Management, January/February 2002, Volume 22, Number 1.