Data Analysis Series: What is Data Structure?

Consider a scenario where you are to walk into a library. The first thing you notice is that every book is arranged according to subject, that is, they are arranged into fiction, nonfiction, or science sections for example. Because there is a catalogue system used by the librarian, you can locate a book easily. But what happens if the system is not there? Finding a single book could take hours, maybe even longer. Just as libraries depend on organization for their effective running, so is data. This is precisely where the idea of a data structure comes into play.

Data structure, or the organization of data is similar to the filing cabinet in the physical world, or even the business world. It sorts, arranges, and stores data in a certain format to enhance access and usage. Without it, data would be a complete mess, hard to find, and almost impossible to study.

Structured Data vs Unstructured Data

Think of our librarian, for instance. She categorizes books first by type, then by author, and finally by title. This is a systematic arrangement that enables readers to find exactly what they need. Similarly, structured data is organized in a standardized format, such as tables with rows and columns. 

Rows represent individual observations, essentially different books in the library, whereas columns describe characteristics of those observations such as title, author, and genre. Structured data is held in a relational database such as MySQL, Microsoft SQL Server, or even a spreadsheet such as Google Sheets or Microsoft Excel.

Now, imagine a store full of old journals, photo albums, and letters. The information is valuable, but it’s like trying to find a black cat in that dark storeroom. This is what unstructured data looks like: it doesn’t conform to a predetermined format, which makes its storage and analysis a bit more challenging. Yet, most information businesses get comes from unstructured data, including emails, social media comments, videos, and images.

It is only with advanced techniques like data analysis skills such as text mining, machine learning, and Natural Language Processing (NLP) that unstructured data can be sorted out. These approaches organize and analyze unstructured data that has been stored. Specialized tools are being built to store unstructured data such as NoSQL databases like MongoDB and Cassandra.

How Data Structure Enhances Analysis

Aspiring data analysts often start by practising with small datasets, which are collections of data organized in a tabular format. These datasets, available from sources like Kaggle, government portals, and Google Datasets, offer a playground for honing analytical skills.

Recognizing the type of data structure being used is crucial before diving into analysis. This would help data analysts pick the right tools to work with. Accessing data stored in a relational database is different from getting information from a NoSQL database. This will guarantee that the analysis is accurate and that inconsistencies, missing values, or errors are identified upfront.

Think of it as the difference between finding a book in a library versus an unorganised storeroom. In the library, you’d use the catalogue. In the storeroom, you might have to sift through boxes with a torchlight. Knowing the environment helps you choose the right tool for the job.

Types of Data Structures

Data structures can be broadly categorized into primitive and non-primitive types:

1. Primitive Data Structures

These are the basic types of data structures that serve as the building blocks for more complex structures. They represent a single value or element.

• Integer: Represents whole numbers.
• Float: Represents decimal numbers.
• Character: Represents individual characters or symbols (e.g., ‘a’, ‘1’).
• Boolean: Represents truth values (True/False).
• String: Represents sequences of characters (though strings are sometimes considered as an array of characters).

2. Non-Primitive Data Structures

These are more complex structures that are composed of multiple primitive data elements. They can be classified as linear or non-linear.

Linear Data Structures

In linear data structures, elements are stored in a sequential manner, and each element has a unique predecessor and successor (except for the first and last element).

• Array: A collection of elements of the same type stored in contiguous memory locations. Access to elements is fast, but resizing can be expensive.
• Linked List: A linear collection of nodes, where each node points to the next node in the sequence. Unlike arrays, linked lists don’t require contiguous memory, which can be advantageous in dynamic memory allocation scenarios. Types include:
• Stack: A collection of elements with “Last In, First Out” (LIFO) access order. Operations are typically restricted to push (add element) and pop (remove element).
• Queue: A collection of elements with “First In, First Out” (FIFO) access order. Common operations are enqueue (insert) and dequeue (remove).

Non-Linear Data Structures

In non-linear data structures, elements do not have a unique predecessor or successor. They are typically used for more complex relationships.

• Tree: A hierarchical structure with nodes connected by edges. The topmost node is called the root, and each node can have zero or more child nodes. Types of trees include:
• Graph: A set of nodes (vertices) connected by edges. Graphs can be used to represent networks like social media connections, routing algorithms, etc. They can be implemented using adjacency matrices or adjacency lists.

The Power of Data Structure

Just as a librarian converts a bunch of loose books into a treasure trove of information by organizing them properly, data structures turn raw information into actionable insights. If an organization identifies and leverages the right structure, it will unlock its data potential, make more informed decisions, and unlock key trends.

In the ever-changing field of data, understanding structure is not only a technical requirement but also the foundation of innovation and discovery.