Naming a value's type isn't guessing. It's two questions, asked in order, and they never fail you.
The Two Questions
Question one — is it a label, or a number? If it sorts things into named groups (red/blue, yes/no, Year 7/Year 8), it's categorical and you're done. If it's a genuine number you'd count or measure, it's numerical — go to question two.
Question two — can it only take whole steps, or any value on a scale? If it jumps in whole counts (3 goals, 2 pets), it's discrete. If it flows smoothly so there's always a value between any two (1.7 m, 22.4 °C), it's continuous.
That's the entire decision: label or number; if number, jumps or flows. Run those two questions and every value lands in exactly one box.
A Dataset Is a Table
Out in the wild you don't get one value, you get a dataset — and a dataset is just a table. Each row is one thing you observed (one student, one day, one frog). Each column is a variable — one feature you recorded about every thing (their height, their favourite sport, the temperature that day). And here's the rule that makes the whole topic tidy: every column has one data type. The “height” column is all continuous numbers; the “eye colour” column is all categorical labels. You don't type single values in real life — you type whole columns.
A Worked One, Slowly
A class survey has these columns: Name, Favourite sport, Number of siblings, Height (cm), Travels by bus? (Y/N). Take them one at a time with the two questions. Name — a label, so categorical. Favourite sport — a label, categorical. Number of siblings — a number, and it jumps in whole counts, so numerical → discrete. Height — a number that flows smoothly, so numerical → continuous. Travels by bus? — yes or no is a label, categorical. Five columns, five clean answers, and now you know exactly what each one will let you do later. Naming the column first is the move that earns marks every time.