Data Normalisation Tool

Introduction

Before feeding columns with different units into a model, or plotting them on one axis, they need a common scale. This page accepts a pasted block of numbers and offers Min-Max, Z-score, decimal scaling and Max-Abs; the selected method prints every original value beside its transformed result.

The example uses 12, 18, 25, 31 and 42 with Min-Max onto 0 to 1. The minimum is 12, the maximum 42, the mean 25.6 and the population standard deviation about 10.40. Twelve maps to 0, 42 maps to 1, and 25 lands near 0.4333. On Z-score that same 25 becomes about -0.0577.

Newlines, ASCII or full-width commas, Chinese enumeration marks, semicolons, spaces and tabs all work as separators, and any unparseable fragment rejects the whole set, so a pasted spreadsheet column can never silently lose rows.

Key features

• Four methods: Min-Max, Z-score, decimal scaling and Max-Abs are all available from one dropdown.
• Row by row: Every original value is paired with its normalised result in the breakdown.
• Custom range: Min-Max bounds are yours to set. They are not locked to 0 and 1.
• Mixed separators: Newlines, ASCII and full-width commas, enumeration marks, semicolons and spaces all split values.
• All or nothing: One unparseable fragment rejects the whole set. Samples are never silently lost.
• Degenerate input: Identical values make Min-Max and Z-score undefined, and both are blocked with a message.

How to use it

1. Raw data: Paste a series of numbers. Commas, full-width commas, enumeration marks, semicolons, newlines and spaces all separate values. An unreadable fragment rejects the whole set rather than quietly dropping a sample.
2. Method: Choose Min-Max, Z-score, decimal scaling or Max-Abs first, then see which of the boxes below it actually needs.
3. Range low, Range high: Only Min-Max uses these. The other three methods work with the bounds left blank.
4. Decimal places: Set how many digits to show. It changes display only, not the arithmetic behind it.
5. Read the breakdown: Press Calculate for sample count, extremes and mean in the first six cells, then read the per-item normalised values row by row below.

Things to keep in mind

• Separators: Newlines, ASCII and full-width commas, enumeration marks, semicolons and spaces all split values.
• Unreadable fragments: Any unparseable fragment rejects the whole set instead of being skipped.
• Identical values: Identical values make Min-Max and Z-score undefined, and both are blocked.
• Unused bounds: Z-score and Max-Abs do not need target range bounds.
• Standard deviation: Population standard deviation uses n; convert yourself if you need sample.

Min-Max needs a spread: when every value is identical the denominator is zero and the method is refused rather than returning a row of blanks. Decimal scaling chooses its power of ten from the largest magnitude in the set, so adding one much larger value changes the scale for every row. The sample count in the first cell is the easiest check that nothing was dropped while parsing.

Paste 1,2,3，4,5 with a full-width comma in the middle and all five values are still counted; the parser treats both comma forms the same way. Add a stray letter anywhere and the whole set is refused instead of being quietly reduced to four numbers. The sample count in the first cell is the quickest confirmation that every value survived.