Skip to content
Tutorials Center
My learning

Topics

Lesson 2 of 76 ยท What Data Science Is

Data Science vs Data Analytics, Applications and Roles

Compare data science, data analytics, business intelligence and machine learning engineering, see the four kinds of questions each answers, and tour real applications by industry.

People use "data science", "data analytics", "business intelligence" and "machine learning" as if they meant the same thing. Job ads mix them, and the boundaries really are blurry. They are still different kinds of work, with different questions, tools and outputs, and knowing which one you are being asked to do saves a lot of confusion.

This lesson draws the map. You will see the four kinds of questions you can ask of data, how analytics, data science, BI and ML engineering split them up, which roles exist around a data team, and what the work looks like in finance, retail, health, transport, media and agriculture. If you have not yet read What Is Data Science?, do that first; this page builds on it.

Four kinds of questions

Almost every data question falls into one of four groups, each harder and more valuable than the last. Analytics mostly lives in the first two, data science extends into the last two.

Four questions you can ask of data

Descriptive What happened?
Diagnostic Why did it happen?
Predictive What will happen?
Prescriptive What should we do?

Each step needs more data skill and more care.

1 / 2

The four types are not rivals. A good project often walks through all of them: describe last quarter's sales, diagnose the dip in March, predict next quarter and recommend a plan. Here is the same small business answered at each level, using code you already know from descriptive statistics.

Python Descriptive and diagnostic questions

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Notice the last two lines: promo months look better on average, but revenue was also growing all year. Comparing change from the previous month is one small step toward a fair comparison. The full story belongs to causal thinking.

Python Predictive and prescriptive questions

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The prediction is only as good as the assumption behind it (a steady straight line), and the recommendation is only as good as the guessed effects of the options. Prescriptive work always rests on assumptions, so a data scientist states them out loud.

Data science vs data analytics

The two fields share tools (SQL, Python, spreadsheets, charts), but they differ in focus.

Data analytics Data science
Main question What happened and why? What will happen, what should we do, how do we automate it?
Typical output Dashboards, reports, ad hoc answers Models, experiments, prototypes, data products
Time horizon The past and present The future
Maths depth Descriptive statistics, simple tests Probability, modelling, optimisation, sometimes machine learning
Typical tools SQL, Excel, BI tools, pandas Python, scikit-learn, notebooks, statistics packages
Typical scale Existing tables Messy, large or unstructured data

Where BI and ML engineering fit

Business intelligence (BI) is the practice of turning company data into regularly refreshed dashboards that managers read, built in tools such as Power BI or Tableau. It is analytics made repeatable. See Data Science with Excel and BI Tools.

Machine learning engineering takes a model that works in a notebook and makes it work in the real world: packaging, serving, monitoring, retraining and keeping it fast and cheap. It is software engineering with models inside. The link between the two worlds is covered in From Model to Decision.

flowchart LR
    A["Raw data"] --> B["Data engineer<br/>pipelines and storage"]
    B --> C["Analyst<br/>reports and dashboards"]
    B --> D["Data scientist<br/>models and experiments"]
    D --> E["ML engineer<br/>deploy and monitor"]
    C --> F["Decision"]
    E --> F

Roles around the data flow. In a small company one person does several of these jobs.

The roles around a data team

  • Data analyst: answers business questions with SQL, spreadsheets and charts; owns dashboards.
  • Data scientist: frames problems, designs experiments, builds and evaluates models.
  • Data engineer: builds the pipelines and warehouses that deliver clean, reliable data.
  • ML engineer: turns models into production services.
  • Analytics engineer: writes tested, documented transformations in SQL that analysts build on.
  • Statistician or researcher: designs studies and tests with rigorous inference.
  • Product analyst or manager: connects the analysis to product decisions.

A beginner usually starts at the analyst end and moves toward modelling, because the early skills (SQL, cleaning, charts, honest statistics) are the foundation of everything else.

Applications by industry

Where data science is used

Finance Fraud, credit risk, pricing
Retail Demand, recommendations
Health Risk scores, imaging
Transport Routes, arrival times

Same workflow, different data.

1 / 2
  • Finance: flag fraudulent card payments in milliseconds, estimate the chance that a borrower defaults, detect money laundering, optimise a portfolio. Mistakes are costly, so models are audited and explained; see Model Interpretation.
  • Retail and e-commerce: forecast demand per store, set prices, recommend products, segment customers (see Clustering and Segmentation) and find items that sell together (Association Rules).
  • Health care: predict readmission risk, support diagnosis from images, plan staffing, analyse clinical trials. Privacy and fairness matter most here (Ethics and Fairness).
  • Transport and logistics: estimate arrival times, plan routes, forecast demand for rides, schedule maintenance, balance a vehicle fleet.
  • Media and entertainment: recommend what to watch or read next, predict which subscribers will cancel (Churn project), measure what content works.
  • Agriculture: forecast crop yield from weather and soil, detect disease from photos, plan irrigation, price a harvest. Data is sparse and seasonal, so time series skills matter.
  • Manufacturing and energy: predict equipment failure from sensor streams, detect defects, forecast electricity demand.
  • Public sector and education: allocate resources, evaluate programs, spot students who need support.

The workflow in The Data Science Workflow is the same in every industry; the domain knowledge you add changes. A model that is brilliant on tables but ignores how a hospital actually works will not be used.

When a job title or a project brief is unclear, ask which of the four questions is actually being asked. It tells you whether you need a dashboard, an experiment, a forecast or a production system.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. A manager asks, 'Why did sales drop in March?' Which type of question is this?

    Diagnostic questions look for causes of something that already happened. Descriptive would be 'how much did sales drop?'
  2. 2. Which role mainly keeps models running reliably in production, with monitoring and retraining?

    ML engineers take a model from a notebook to a service that is deployed, monitored and updated.
  3. 3. What is business intelligence mainly about?

    BI turns data into regularly refreshed dashboards and reports that decision makers read. It is analytics made repeatable.
  4. 4. Which statement about the four question types is true?

    Many projects describe, diagnose, predict and then recommend. Prescriptive answers always rest on stated assumptions.

Summary

  • Data can answer four kinds of questions: descriptive, diagnostic, predictive and prescriptive.
  • Analytics focuses on the past and present; data science adds prediction, experiments and automation.
  • BI makes analytics repeatable; ML engineering puts models into production; data engineers feed both.
  • Roles overlap, especially in small teams, so know the question before choosing the tool.
  • Applications span finance, retail, health, transport, media, agriculture, industry and the public sector, all following the same workflow.

Next: The Data Science Workflow.