written by Maurizio Crisanti
Data modeling is an increasingly important activity in Data Analysis and Business Intelligence fields. More and more, data has been representing an important value for every company, and it is necessary to analyze it profitably. The design of a Data Model is an indispensable activity to manage data sets, process them and obtain qualified information for the company. Let’s try to understand what a Data model is, how it is made and which its applications are. We will try to answer these questions in this article.
What is a Data Model
Creating a Data Model consists in data processing activities, identification of the logical relationships between them and data construction in graphical format. In essence, creating a Data Model transforms data into accurate and personalized information. These information are essential for detecting in real time the status of markets, sales or measurable phenomena and supporting strategic decisions with a scientific approach. The Data Modeling activity consists in defining the objectives of the model and the relationships between the data, also allowing you to identify any missing or redundant data and avoid producing incorrect reports. The design of a Data Model requires specific skills and time for development, but in the long run, the maintenance and updating of the model are quick and economical.
Phases of realization of a data model
There are mainly three phases in the realization of a Data Model:
- Conceptual scheme: in this phase, generally through a diagram, it should be created a data schema, starting from the reality to be analyzed. In this phase, it is defined which datasets to include in the model and which relationships to detect in collaboration with top managers. The main purpose of this phase is to define the valuable data for the company and the objectives of the model.
- Logical schema: defines how to implement the system and adds additional information to the elements of the conceptual model. In this phase, it is defined the structure of the data elements and the relationships between them. It provides the elements that will be the basis of the physical model.
- Physical schema: consists in the physical realization of a data model using a Database Management System. In this phase, there is the development of the database and the system of elaboration. Moreover, there is creation of the report and of data visualization systems.
Creating a data model: the most used tools
There are many tools that allow you to speed up the creation of a Data Model. Here are the most used.
Erwin Data Modeler (Dm)
It is a very popular data modeling tool. Erwin Data Modeler (DM) allows you to manage a wide range of complex data structures. Conceptual, logical, or physical data models can be created and displayed using an optimized graphical interface. It is a versatile tool with very useful features, such as user collaboration and authorization levels, change tracking, and display of a wide range of roles and relationships between data. The tool interfaces with most databases, including AWS, Azure, Hadoop, Oracle, Teradata, ValidDB and many others.
Magicdraw
Developed by NoMagic Inc., it is a modeling tool designed for a wide variety of modeling languages, including UML, SysML and AADL, and programming (Java, C #, C ++). The tool supports multiple platforms, operating systems and environments, as it was designed as a Java application.
MagicDraw offers group collaboration on a common server, free support, and a rather intuitive design. The developers of the platform state that 75% of the added features are based on direct user feedback. There is a standard, a professional and an enterprise version.
Powerdesigner
PowerDesigner 16.6 from SAP is a professional Data Modeling tool, used in database design in Windows and Eclipse environments. It offers conceptual, logical and physical models, UML diagrams, Java J2EE, Microsoft .NET, Visual Studio and much more. Some of the main features supported include data visualization, impact analysis and search and reuse and other features mentioned in the above software.
Argouml
Here is a completely free and open-source data modeling tool: ArgoUML offers a standardized interface for viewing all UML files. Like MagicDraw, the tool is built in Java language, which makes it compatible with all platforms. The tool requires only 15MB of free disk space and it is the lightest database design tool on the market. The platform also hosts a wiki, an online forum and an in-depth user manual that facilitate its use.
The applications of a Data Model
The applications of a Data Model cover all sectors. If well designed, the model is able to provide the company management with useful elements to guide data-driven decisions. In the most advanced versions, a Data Platform integrated by AI algorithms, such as DataLysm, can provide elements of predictive analysis, already widely used in the weather forecasting sector, in the finance sector and in digital marketing.
DataLysm is a platform developed by 3rdPlace for its customers. Here are some applications:
Insurance Sector
Thanks to a Data Modeling activity, it was possible to analytically measure the risk appetite of motorists, and then redefine insurance premiums. The data collected from the blackboxes installed on the cars were inserted into a Data Model with Artificial Intelligence.
Business Information & Credit Management Sector
The client requested the best measurement of the companies’ default risk. It was decided to enrich traditional datasets, such as economic and financial data, with digital data, for example the activities and reputational analysis on the web of the individual company, by applying AI to the model.
Consumer Goods Sector
The need was to measure the potential and performance of physical cross-country stores, by identifying areas of potential buyers. It was decided to segment the users of one or more points of sale within specific markets through the combination of territorial socio-economic data, distribution characteristics and Alternative Data. It was therefore possible to create a score to facilitate the alignment of some stores to target stores.
Conclusions
Creating a Data Model allows you to organize and analyze data, creating easily intelligible reports, available to all company sectors. It is possible to relate company data with external data sets, automating the creation of graphic representations that accurately photograph reality, detecting it dynamically.
A data model makes it possible to extract economic and strategic value to databases, showing market trends, the results of marketing operations and insights, offering valuable information, before competitors.
