Data Concepts and Environments
This chapter establishes the vocabulary of data analytics. You will learn the types of data, the systems that store it, the difference between transactional and analytical processing, and how warehouses, lakes, and marts differ.
Types of data
Data is commonly classified by structure. Structured data fits neatly into fixed rows and columns, as in a relational table. Unstructured data, such as images, video, and free text, has no predefined model. Semi-structured data, such as JSON and XML, carries organizing markers like tags or keys but does not conform to a rigid table schema. Recognizing the type guides how data can be stored, queried, and analyzed.
Databases and storage systems
Relational databases organize data into related tables with enforced schemas, while NoSQL databases offer flexible models for large or varied data. For analytics, a data warehouse consolidates cleaned, structured data for reporting, a data lake stores large volumes of raw data in native format for later processing, and a data mart is a focused subset serving a specific department or subject area.
OLTP versus OLAP
Online transaction processing (OLTP) systems are optimized for many small, fast reads and writes from operational applications, such as recording individual sales. Online analytical processing (OLAP) systems are optimized for complex queries that aggregate large volumes of historical data for analysis. Understanding this distinction helps analysts know where to run heavy analytical queries without harming operational performance.
Dimensions and measures
In analytical modeling, measures are the numeric values you analyze, such as revenue or quantity, while dimensions are the descriptive attributes you slice by, such as date, region, or product category. Fact tables hold measures and reference dimension tables, forming the star schema that underlies much reporting and makes queries efficient and intuitive.