Evolutive Data Model

Introduction

In today's era of digital and increasingly AI driven era, data stands out as the vital new core of financial and banking institutions.

Data integrates information across all layers and provides a powerful foundation for extension, evolution and even revolution.

What is data modeling?

Data modeling is the process of creating a data model for the data to be stored in a specific format in a database.

Data models identify what data is needed and how it should be organized, enabling the creation of a database that will be used to develop an app. They also ensure the quality of data via naming conventions and default values.

Data models can serve a variety of purposes, from high-level conceptual and logical models to physical data models (PDMs). We’ll briefly describe the types of data models in the next section.

Types of data models

Developing an active database requires creating three main data models:

  • Conceptual Model – Identifies and organizes business concepts. Addressing the business requirements, it defines what data the system contains.
  • Logical Model – Defines how the data rules and structures are mapped. Addressing the data requirements, this model is the base for developing the physical mode.
  • Physical Model – Describes how the data is structured in the DB. Addressing the technical and performance requirements, it is the actual structure of the database that will be used to develop an app.

From bottom to top, each data model serves as foundation for the next data model, gradually adding more details about the data and data non-related properties until producing the actual structure of the database.

Data models’ creation implies using specific data modeling techniques. These are the two major data modeling techniques:

  • Entity Relationship Diagram (ERD) – A high-level conceptual data model diagram which provides a visual representation of the data and how it is interconnected.
  • UML (Unified Modeling Language) – A generic data modeling language that standardizes the data, enabling the design of a system. An UML may consist of more than one ERD.

Now that we’ve walked you through the types of data models and data modeling techniques, let’s create data models from bottom to top (conceptual to physical).

Data Modeling

While creating data models from physical to conceptual is useful in reverse engineering to extract models from existing systems, creating data models from conceptual to physical models serves as a powerful template and reference for your DB, enabling stakeholders identify gaps and make proper changes before programming an app.

Entity Relationship (ER) modeling is a best practice for producing well-designed databases. It depicts the structure of a relational databases allowing you to understand the data and how it shares information.

The main concepts of an ER data model are:

  • Entities – objects representing things from the real world. An entity is a collection of items having a common set of attributes, and each item has a combination of attribute values.
  • Attributes – characteristics of properties of the entities.
  • Relationships – how entities are related one to another; how they share information in the DB.

Entities and Attributes

An entity is an object representing a thing from the real world. Entities are tables in a database (DB), each column representing an attribute which stores the value of an entity characteristic. Attributes have specific properties based on the data type.

Relationships

Relationships define how entities share information in the database. An important aspect of relationships is the cardinality; it defines how data is related between the entities: none, one-to-many many-to-many. Cardinality defines how records of an entity are related to the records of another entity. For example, multiple customers can have the same account type.

This section describes how to create data models from conceptual to physical models using the ER model.

STEP 1. Create Contextual Data Model

Identify the entities, their attributes and the relationships between entities. The contextual data model does not provide specific details of the relationships nor details of the actual database structure.

STEP 2. Create Logical Data Model

This model adds additional information to entities and attributes identified in the contextual data model.

An entity is described by at least two attributes: a primary key which uniquely identifies an entity and at least another attribute which provides an entity characteristic. For example, the primary key of the Customer entity is the AccountId attribute. It uniquely identifies the records of the Customer entity.

To create the logical data model starting from a contextual data model, for each entity define which attribute is the primary key and for all other attributes, define their type (text, numeric, date, whole number, etc.) and properties (length, required level, etc.).

Although the DB structure is still generic, the logical data model provides the baseline for the physical data model.

STEP 3. Create Physical Data Model

Once you’ve defined the contextual and logical data models, you can create the actual database structure by defining the relations between entities.

An important aspect of entity relationships is cardinality. Cardinality is the property of the relationship itself, specifying how records of an entity are related to the records of another entity: one-to-many or many-to-many.

To establishes a link between records of two different entities (entity relationship), you have to add an attribute of type lookup, also known as foreign key.\

When talking about relationships, we distinguish two entities: parent and child. The entity to which you add the foreign key becomes parent entity for the related entity also known as child entity.

Below is an example of a basic physical data model created in FintechOS Studio, comprised of two entities, their attributes and relationship:

 

You can extend the physical model with new attributes or with data from external systems, creating an evolutive data model.

Evolutive Data Models

Data models are the basis for building evolutive data models. Evolutive data models not only ensure the modeling of the database structure, but they also extract data from legacy systems, processes and data repositories, extending data by combining and connecting data.

FintechOS Studio provides you with various options to gather and interconnect data. To mention a few: REST API, Data Pipes and REST Pipes.