A carpenter doesn't use a hammer for everything — they reach for the right tool for each job. Data scientists are the same: a spreadsheet for a quick look, SQL to pull data from a database, Python to do the heavy lifting, and Git to keep it all safe. Knowing which tool does which job means you'll never feel lost opening a real project. By the end of this lesson you'll know the whole toolkit and where each piece fits.
What you'll learn
What
The core data science toolkit is a small set of tools: spreadsheets for quick exploration, SQL to get data out of databases, Python (in a Jupyter notebook) as the main workhorse, and Git for version control. Each maps to stages of the workflow you just learned.
Why
There are hundreds of data tools, and it's easy to feel overwhelmed. In reality a handful cover the vast majority of daily work. Knowing what each is for — and when to reach for it — saves you from learning everything at once.
Where it's used
Every data team on earth uses some mix of these. Job postings for data roles list Python, SQL, and Git so consistently that they're effectively the entry ticket to the field.
Where this runs in production
Analysts write SQL to extract billions of viewing records from data warehouses, then switch to Python notebooks to explore and model them. The two tools hand off exactly where the workflow expects.
Google built Colab, a free hosted Jupyter notebook, because notebooks are how data scientists actually work — write a little code, see the result, adjust, repeat.
Data teams store notebooks and scripts in Git so every analysis has a history: who changed what, when, and the ability to roll back a mistake. It's the safety net under the whole workflow.
Spreadsheet — quick look at small data. SQL — the language for getting data out of databases. Python — the main programming language for cleaning, analysis, and modelling. Jupyter notebook — the interactive workspace where you write and run Python step by step. Git — version control that tracks every change and lets you undo mistakes.
A database is a huge kitchen full of ingredients. You don't wander in and grab things — you hand the kitchen an order: 'all orders from customers in Kenya, last month, just the date and total.' SQL is the language of that order slip. The kitchen (the database) does the fetching and hands back exactly what you asked for. The analogy breaks down in that, unlike a kitchen, the database returns results in milliseconds even from billions of rows.
# A Jupyter notebook cell — write a little, run it, see the result revenue = 4200 units = 120 print(revenue / units) # 35.0 <- output appears right below the cell
A Jupyter notebook lets you run code in small chunks called cells and see each result immediately underneath. That tight write-run-see loop is why notebooks are the default home for exploration — you think and experiment in the same place, unlike a plain script you run all at once.
Python is the language; Jupyter is one place you can write it. They're often used together but aren't the same thing — you can run Python without Jupyter, and Jupyter can run other languages too. Beginners often say 'Jupyter' when they mean 'Python.' Keep the two ideas separate.
Click each tool to see its main job, its limits, and the workflow stage it powers.
Select a type to see its full definition, operations, and data science usage.
A colleague emails you a 300-row CSV and asks for the total and a quick bar chart in five minutes.
Small, one-time, no repeatability needed. This is exactly what a spreadsheet is best at — open it, sum a column, insert a chart. Reaching for Python here would be slower.
Tool: spreadsheet (fast, small, one-off).
Your task
Match each job to the right tool by filling in the blanks. Use exactly one of these strings for each: "spreadsheet", "SQL", "Python", "Git". Think about what each tool is best at, then run the code.
spreadsheet SQL Python Git
Write your solution in the editor on the right, then hit Run.
What is SQL primarily used for?
Which tool is the main workhorse for cleaning, exploring, and modelling data?
What does a Jupyter notebook let you do that a plain script does not emphasise?
What's the better approach?
Which statement about Python and Jupyter is correct?
Which tool would have prevented both problems?
What tools would you expect to use day-to-day as a data scientist, and what is each one for?
The core set is small. SQL for getting data out of databases — that's usually how a task starts. Python as the main workhorse for cleaning, exploring, and modelling, using libraries like pandas and scikit-learn. Jupyter notebooks as the interactive workspace where I write and run that Python step by step and see results immediately. Spreadsheets like Excel or Google Sheets for a quick look at small data or sharing a simple result with non-technical colleagues. And Git for version control, so my work is saved, reproducible, and shareable. Each maps onto the workflow: SQL for Collect, Python-in-Jupyter for Clean, Explore, and Model, and Git spanning everything as a safety net.
When would you choose a spreadsheet over Python, and when would you make the switch?
I'd use a spreadsheet when the data is small — a few thousand rows — the task is one-off, and I need a fast answer or a simple chart to share with non-technical people. Its strengths are speed and familiarity. I'd switch to Python when any of three things appear: scale, because spreadsheets get slow and unstable past tens of thousands of rows; repeatability, because a task I'll rerun monthly should be code I can replay exactly rather than clicks I redo by hand; or complexity, like joining multiple sources, serious cleaning, or modelling. The tipping point is usually repeatability and error-risk — the moment a manual spreadsheet process becomes something people rely on, the lack of reproducibility and the ease of a silent copy-paste error make code the safer choice.
Why does version control matter for data science specifically, when the analysis 'works' on your machine?
'Works on my machine' is exactly the problem version control solves. Analyses evolve through many iterations, and without Git you can't reliably answer which version of the code produced a given result — which is a serious issue when a number ends up in a business decision or a published figure. Git gives you recoverable snapshots, so a dead laptop or a bad edit doesn't erase a week of work, and a full history so a teammate can see who changed what and why, and reproduce last month's output precisely. On a team it also lets several people work on the same project without overwriting each other. The deeper point is reproducibility: good data science should be re-runnable by someone else and give the same answer, and version control is the foundation that makes that possible. It's less critical for a genuine one-off, but anything you'll revisit or share belongs in Git.
Common Mistakes to Avoid
1) Forcing a spreadsheet to do work it's outgrown — huge datasets or repeatable pipelines belong in code. 2) Confusing Python (the language) with Jupyter (one workspace for it). 3) Thinking SQL can model or plot data — its job is fetching from databases. 4) Skipping version control on anything you'll revisit or share, then losing work or forgetting which version produced a result.
Ask the AI Tutor
Try these prompts in the AI Tutor panel: • 'Give me a task and quiz me on which tool I'd use.' • 'ELI5: what is version control?' • 'Explain the difference between Python and Jupyter with an example.' • 'Show me a tiny SQL query and explain each part.' • 'Interview mode: ask me which tools I'd use for a real project and why.'
Glossary
Spreadsheet — a grid tool (Excel, Google Sheets) for quick looks at small data. SQL (Structured Query Language) — the language for getting data out of databases. Python — the main general-purpose language for data cleaning, analysis, and modelling. Library — a reusable bundle of code (e.g. pandas, scikit-learn) that adds capabilities to Python. Jupyter notebook — an interactive workspace that runs code in cells and shows results inline. Git — a version control system that snapshots your work. GitHub — a website for hosting and sharing Git projects.
Recommended Resources
• Try it free: Google Colab (colab.research.google.com) — a hosted Jupyter notebook you can open in a browser with no setup. • Reference: the official pandas '10 minutes to pandas' guide, for when you reach the Python domain. • Overview: 'Git and GitHub for Beginners' introductory videos on YouTube. • Next in DSM: this completes the 'What Is Data Science?' module. Up next you'll move from knowing about data to working with it — reading and interpreting real datasets.
Recap
✓ The core toolkit is five tools: spreadsheet, SQL, Python, Jupyter, and Git. ✓ Spreadsheets suit quick looks at small data; SQL gets data out of databases (Collect). ✓ Python is the workhorse for cleaning, exploring, and modelling; Jupyter is where you run it interactively. ✓ Python is a language and Jupyter is a workspace — related, but not the same thing. ✓ Git is version control: it saves your work, lets you undo mistakes, and makes analysis reproducible. That completes the 'What Is Data Science?' module. You now know what the field is, the kinds of data it uses, the workflow every project follows, and the tools that power it. Next module — Understanding Data — you'll stop reading about data and start reading data itself.
Run your code to see the output here.