What is data modeling?
Data modeling is a process for defining and ordering data for use and analysis by certain business processes. The goal of data modeling is to produce high quality, consistent, structured data for running business applications and achieving consistent results. Data modeling is influenced by the type of IT systems used and the objectives of the organization involved. Data modeling encompasses three progressive steps:
- Conceptual: to define the model’s parameters;
- Logical: to structure the data for use; and,
- Physical: to produce sets of data in a common format for analysis and use by the organization.
Data Model Examples
There are two primary approaches to data modeling: top-down and bottom-up.
- Top-down data modeling starts with a question and works to arrive at an answer. For example, an organization may believe that certain inefficiencies are costing it money, and so it gathers data from its various departments to analyze resource utilization and identify ways to eliminate redundancies and better allocate available capital.
- Bottom-up data modeling starts with the collection of data and seeks to find interesting insights and correlations. For example, by running all of its transactional data, an organization might learn that customers are more likely to respond.
Top-down and bottom-up data modeling are complementary. Bottom-up modeling may reveal questions that top-down modeling can then answer; and top-down modeling may be used to identify different sources and combinations of data for analysis.
Data modeling has been used to achieve better outcomes in nearly every industry, including:
- Business management
- Civil engineering
- Environmental science
- Financial services
- IT management
- Research and development
- Software development
- Transportation and logistics
What are some data modeling best practices?
Because data modeling has so many applications, but may be difficult for some to understand, it is recommended that teams tasked with data modeling for their organization gain buy-in from the groups they’ll be working with by communicating—in non-technical terms—the benefits of cooperation. Once buy-in has been achieved, focus on simpler tasks first to demonstrate success and cultivate champions. Other best practices to follow include:
- Outline processes and objectives in advance
- Document each step of your process to facilitate course correction and ensure replicability
- Work collaboratively with all associated teams in order to leverage domain expertise
- Revisit your model regularly and modify based on updated objectives and data sources
Why is having a common data model important?
In a large, complex enterprise, there are many and diverse sources of structured, semi-structured, and unstructured data. All of it may be useful for understanding operations, but if the formats are inconsistent the application of the data will produce inconsistent and unreliable results. Data modeling transforms diverse data sets into a common format, maximizing its usefulness to the business processes to which it is being applied.« Back to Glossary Index