Data & AI
Data Analysis
Data analysts answer business questions with evidence. The technical skills are learnable in months; what makes an analyst valuable is asking the right question and communicating the answer to someone who will act on it.
Start here
Download any public dataset that interests you and open it in a spreadsheet. Ask it one question. That is the job. Start at Stage 1 with no prior experience.
Foundations
The two tools that do most of the work, and the thinking that makes them useful.
RequiredSpreadsheets, properly
Dismissed by people learning to code, and still the most-used analysis tool in almost every organisation you will work for.
Spreadsheets, properly
Dismissed by people learning to code, and still the most-used analysis tool in almost every organisation you will work for.
What to learn
- Pivot tables
- Lookup functions
- Cleaning imported data
- Charts that do not mislead
- Knowing when a spreadsheet is the wrong tool
Practice
Next: SQL, which is the single most valuable skill on this path.
RequiredSQL
The highest-return skill in data analysis by a wide margin. Almost every analyst job description requires it and almost every interview tests it.
SQL
The highest-return skill in data analysis by a wide margin. Almost every analyst job description requires it and almost every interview tests it.
What to learn
- SELECT, WHERE, GROUP BY
- JOINs and when each applies
- Subqueries and CTEs
- Window functions
- Aggregations and NULL behaviour
Tools
- PostgreSQL
- A SQL client
Practice
Project
beginnerA SQL analysis
Load a public dataset into a database and answer five specific questions with queries, writing one paragraph on what each answer means.
- PostgreSQL
- A public dataset
You can go from a raw dataset to a defensible answer without leaving SQL.
Next: Statistics, so the answers are honest.
Analytical thinking
Where analysts go wrong is almost never the query. It is the interpretation.
RequiredPractical statistics
Enough to avoid confidently reporting something untrue, which is the fastest way to lose an organisation's trust.
Practical statistics
Enough to avoid confidently reporting something untrue, which is the fastest way to lose an organisation's trust.
What to learn
- Distributions and summary statistics
- Sampling and bias
- Correlation versus causation
- Significance and confidence in plain terms
- Simpson's paradox and other traps
Practice
Next: Cleaning data, which is most of the actual work.
RequiredData cleaning and validation
Realistically the majority of the job. Nobody hands you clean data, ever.
Data cleaning and validation
Realistically the majority of the job. Nobody hands you clean data, ever.
What to learn
- Missing values and what to do about them
- Duplicates and deduplication logic
- Inconsistent categories and dates
- Outliers versus errors
- Documenting the assumptions you made
Next: Python, once spreadsheets stop scaling.
Analysis in code
The point at which your work becomes repeatable rather than a sequence of clicks.
RequiredPython for analysis
A spreadsheet analysis cannot be re-run next month without redoing it. A script can.
Python for analysis
A spreadsheet analysis cannot be re-run next month without redoing it. A script can.
What to learn
- Python basics
- pandas for tabular data
- Reading from files, APIs and databases
- Grouping and joining in code
- Notebooks and their pitfalls
Tools
- Python
- pandas
- Jupyter
Practice
Next: Visualisation.
RequiredVisualisation
A correct analysis nobody understands has not achieved anything.
Visualisation
A correct analysis nobody understands has not achieved anything.
What to learn
- Choosing the right chart for the question
- Avoiding misleading axes
- Colour and accessibility
- Annotating the point rather than leaving it to be found
- Restraint
Tools
- matplotlib or seaborn
- Power BI or Tableau
Practice
Next: Dashboards.
Delivering insight
Where analysts become valuable rather than merely competent.
RequiredBusiness intelligence tools
Most organisations expect self-service dashboards, and building ones people actually use is a distinct skill.
Business intelligence tools
Most organisations expect self-service dashboards, and building ones people actually use is a distinct skill.
What to learn
- Data modelling for reporting
- Building dashboards for a specific decision
- Filters and drill-down
- Refresh and data freshness
- Avoiding dashboards nobody opens
Tools
- Power BI
- Tableau
- Looker Studio
Project
intermediateA decision-focused dashboard
A dashboard built for one named decision a specific person has to make, with the data pipeline behind it automated and documented.
- Power BI or Tableau
- SQL
Someone other than you uses it regularly, and you know which decision it supports.
Next: Communication.
RequiredCommunicating findings
The difference between an analyst who is consulted and one who is sent requests.
Communicating findings
The difference between an analyst who is consulted and one who is sent requests.
What to learn
- Leading with the answer
- Writing for an executive reader
- Quantifying uncertainty without hedging everything
- Recommending an action
- Handling challenges to your numbers
Project
advancedAn end-to-end analysis with a recommendation
Take a real question, gather and clean the data, analyse it, and produce a short written report that opens with a recommendation and defends it. Include what would change your mind.
- Python
- SQL
- A visualisation tool
You have produced something a decision-maker could act on without asking you to explain it first.
Where this leads
You do not have to pick one now. These are the directions this path opens up once you are working.
Continue your journey
The paths closest to this one. Skills overlap more than the job titles suggest.
Data Engineering
Build the pipelines that make everyone else's data work possible.
ViewData Science
Use statistics and modelling to answer questions data alone cannot.
ViewProduct Management
Decide what gets built, why, and what gets cut.
ViewAI Application Development
Build products people use, that happen to have a model inside.
ViewYou do not have to do this alone
Our programs are free, taught live, and built around the same progression. Join one and work through it alongside other people.