Data Analysis with Python (pandas, polars, DuckDB)
Aspiring data analysts, students, and professionals moving from spreadsheets to code.
This page needs a newer browser to run its tools. You can still read everything here.
Working with data: asking a clear question, getting the data that answers it into shape, and reading the result honestly.
Aspiring data analysts, students, and professionals moving from spreadsheets to code.
Back-end developers and DBAs who run PostgreSQL in production.
Back-end developers, DBAs and SREs who run MySQL or MariaDB: InnoDB, tuning, migrations, replication, HA, backups, upgrades, MariaDB, managed MySQL and Vitess.
Python and JavaScript learners, analysts and developers who need data from the web or want to automate their own web and office tasks, legally and politely.
Developers and analysts moving into pipelines, warehouses and streaming.
Back-end and mobile developers choosing and modelling non-relational databases.
Python programmers moving into data science, ML, engineering, physics, finance or bioinformatics, engineering and science students (including MATLAB users switching to Python), and analysts who outgrow pandas-only code.
Students and professionals in geography, civil and environmental engineering, planning, agriculture, public health and logistics, analysts and developers adding location to their work, and geomatics exam candidates.
Analysis goes wrong most often before any code runs: the question is vague, or the data cannot answer it. Write the decision the analysis should support in one sentence, then list the columns you would need to make it. If a column is missing, or measured differently from what you assumed, you have found the real work.
Look at the rows, not only the totals. Count the missing values, the duplicates and the impossible ones, such as negative ages or dates in the future, and decide what each one means before you drop it.
Keep every cleaning step in code, so you can run it again when the data changes and show someone else exactly what you did.
A chart or a summary should answer the question you started with. Say how many rows it rests on, what you left out and why, and resist reading a cause into a pattern that may be chance.
Prefer a few clear charts to many: label each axis with its unit, start bars at zero, and put the comparison that matters side by side.