Happy New Year 🥂
I'm starting 2026 with a quiet but important realization: A lot of data problems aren't caused by analysis, they're caused by what we do before anyone ever sees the result.
Today was a 4+ hour deep work day, and almost everything I learned sat upstream of insight.
In Data Visualization, I didn't just "plot charts."
I learned how to prepare figures for real humans:
Why PNG is best for reports and dashboards (lossless, clean)
Why JPG works for the web but quietly sacrifices detail
Why SVG matters when designs need to be edited later
How dpi, figure size, and quality can change how trustworthy a chart feels
How to automate figure generation from data using: loop variables and .unique() method to create multiple plots from one dataset.
This taught me that sharing data is a responsibility, not a final step.
Then, in Importing Data, I stepped into relational databases:
What a relational database actually is
How to create a database engine with SQLAlchemy
How data is queried, fetched, and controlled long before analysis begins
And in Cleaning Data, I learned about Data Range Constraints. Data often breaks the rules we set for it.
Ratings exceed their limits. Subscriptions appear in the future. And suddenly you're forced to choose:
Do you drop the data?
Or do you correct it to preserve meaning?
Those aren't technical decisions. They're judgment calls.
That's what tied everything together for me today: Saving figures, querying databases, enforcing data ranges, they all decide whether the story we tell later is honest or misleading.
I wrapped up Data Visualizations with Matplotlib course today, and I'm continuing the remaining two others, with four more interactive courses starting tomorrow.
New year, new standards for how I work with data.
Happy New Year to everyone building carefully, not just quickly.
-SP
Top comments (0)