The Simple Answer
Data Science is the discipline of obtaining knowledge/information from data.
Think of it as a marriage between various tools, algorithms, and machine learning principles that help you find ways to extract hidden knowledge and information from data. It’s like being a detective, only with data.
The Analogy: The Gold Mine
Imagine a big unexplored mountain (this represents the data). The data scientist has the following tasks:
Discover the mountain and get permission to mine (Data Acquisition & Business Understanding).
Dig out the raw ore (Data Collection).
Clean the ore and remove as much dirt and rock as possible (Data Cleaning & Preprocessing). This task alone is often the most time-consuming!
Analyze the ore to see where the vein of gold is located (Exploratory Data Analysis & Model Building).
Extract and refine the gold (Building Predictive Models & Generating Insights).
Make it into a desirable shape like jewelry or coins (Data Visualization & Storytelling) that will help the company decide what to do with the ore.
The Core Components (The Venn Diagram)
The most common way to represent Data Science is by the three-way Venn diagram, where the three main components are:
• Computer Science – the science of large database design, management and algorithms
• Math and Statistics – the understanding of patterns, statistical analysis, probabilities and their representation as algorithms
• Domain Knowledge – the specific problems that one wants to solve using Data Science
While the first two components are obvious, the third one could be anything from healthcare to finance to marketing and even entertainment. The common element in all of them is communication and storytelling that bind together Data Science and its practical implementation in businesses.
The Data Science Lifecycle (The Process)
Data Science projects have a certain set of activities that could be commonly represented as a lifecycle:
1. Problem definition – defining what business problem one wants to solve with Data Science
2. Data acquisition – getting data from various sources such as databases, web scraping or social media
3. Data preparation – pre-processing of data to remove errors or irrelevant information (usually around 80% of data scientists’ time)
4. Exploratory Data Analysis (EDA) and visualization – exploring the data properties and relations to understand what statistical patterns could potentially be inferred
5. Model building – creating a predictive model using machine learning algorithms or other statistical methods
6. Model validation and evaluation – estimating the model’s performance on another set of data to check if it generalizes well
7. Model deployment – releasing the model into production
8. Storytelling, visualization and communication – presenting the result of Data Science research in a compelling and easy-to-grasp way
9. Key Techniques and Tools
Some of the most common techniques and tools in Data Science are:
• Machine Learning – the set of algorithms that either predict or classify patterns in data
• Statistical Analysis – the exploration of data to find properties, relationships and laws
• Programming – mostly Python and R, but other languages could be used
• Data Visualization – the presentation of results and data in graphical form using tools such as Tableau, Power BI or Python libraries
• Big Data – databases that require special processing and tools such as SQL, Spark or Hadoop
Real-World Examples
Some common examples of Data Science in businesses are:
• Recommendation systems – used by Netflix, Spotify or other media-sharing services
• Predictive analytics – for epidemic spread prediction, treatment effectiveness or other health-related data
• Processing of financial transactions – checking credit card fraud
• Market analysis – estimating market size, sales prediction, price optimization or marketing analytics
• Self-driving cars – computer vision could be used for object detection
In a Nutshell
| It’s NOT just… | It IS… |
|---|---|
| Just Statistics | Statistics + Programming + Domain Knowledge |
| Just Coding | Solving business problems with data-driven solutions |
| Just about Building Models | About the entire process, from question to deployment to impact |
| Magic | A rigorous, iterative, and often messy process of discovery |
In essence, data science is a powerful discipline that turns raw data into understanding, actionable insights, and competitive advantage.
