What counts as data
Data is just written-down observations. How to record yours so it is usable later instead of a mess you have to redo.
Data sounds like something that lives in a database, belongs to a company, and arrives in a file.
It is much simpler than that: data is observations, written down consistently. Forty lines in a notebook is data. If you count your bus every morning for two weeks, you have made a dataset — and it is a better one than most files you could download, because you know exactly what every number means.
Consistent is the whole trick
The difference between a dataset and a pile of notes is that a dataset records the same things, the same way, every time.
Bad:
Monday - bus was late
14th, quite late
16 Oct — on time I think
Good:
| date | day | scheduled | actual | minutes_late | |------|-----|-----------|--------|--------------| | 2026-10-13 | Mon | 07:40 | 07:52 | 12 | | 2026-10-14 | Tue | 07:40 | 07:41 | 1 | | 2026-10-16 | Thu | 07:40 | 07:40 | 0 |
Same columns, every row, always filled in. Now you can count, average and compare. The first version cannot be worked with at all — not because it is wrong, but because "quite late" is not a number and Monday is not a date.
Two rules
One row per observation. One row per bus, not one row per week with everything crammed in.
One column per fact. date and minutes_late in separate columns, never
"12 min late on the 13th" in one.
These two rules are the entire idea of a tidy table, and everything you will ever learn about data assumes them.
Record the missing ones
If you miss a day, write the row with the value blank — do not silently skip it.
A skipped day is invisible later, and you will not remember whether the bus did not come, or you overslept, or you were on holiday. Those are three very different facts and only one of them is about buses.
Where to keep it
Your phone's notes app is fine. A free spreadsheet is better, because it will export to CSV, which is the file every tool in the world can read.
Whatever you use, back it up. Losing two weeks of counting to a dead phone is a uniquely miserable way to end a project.
Write down where it came from
Even for your own data, note: what you measured, how, when you started, and anything odd. "The 16th was a holiday, fewer people, bus was early."
Those notes are the difference between a number you can explain in three months and a number you can only apologise for.
Do this now
Set up your table with its column headings before you collect anything. Decide today exactly what each column means and how you will record it.
Deciding halfway through is how datasets become unusable.