Feature Scaling Calculator
Scale raw feature values using min-max normalization, z-score standardization, robust scaling, and max-abs normalization for machine learning preprocessing.
About this calculator
This calculator applies four scaling methods to one Raw Value in parallel, and each formula deliberately ignores the inputs the others depend on: Min-Max Scaled and Max-Abs Normalized never reference Mean or Standard Deviation (lines 11, 15), while Z-Score never references Minimum Value or Maximum Value (line 12) — every method reads only the inputs its own textbook definition calls for. "Robust Scaled Value" here is Raw Value minus Mean, divided by the min-max range (line 13) — a range-based approximation, not the textbook robust scaler, which centers on the median and divides by the interquartile range to resist outliers the way this version cannot. A trap worth flagging explicitly: this calculator ships with Raw Value and Mean both defaulted to 50, so Z-Score, Robust Scaled Value, and Deviations from Mean all read exactly 0 out of the box, and nudging Raw Value, Mean, or even Standard Deviation up and down by the same amount shows zero effect on Deviations from Mean at those defaults — not because those inputs are irrelevant, but because Deviations from Mean is defined as the absolute value of Z-Score (line 23), so a symmetric nudge that pushes the underlying z-score to, say, +0.1 in one direction and −0.1 in the other reads identically once the sign is stripped; move Raw Value to anything other than 50 (breaking that symmetry around zero) and all three respond immediately.
What this does not account for: outlier resistance in "robust" scaling (it uses range, not IQR), and it never checks whether Minimum Value is actually less than Maximum Value before computing a range.
Inputs
Results
Min-Max Scaled Value
0.5
Z-Score (Standard Scaled)
0
How to Use This Calculator
- Enter the feature values or paste summary statistics (min, max, mean, std).
- Select the scaling method: standardization (Z-score), min-max normalization, or robust scaling.
- Review scaled feature range and distribution.
- Use standardization for distance-based algorithms (KNN, SVM) and min-max for neural networks.
- Always fit scaling parameters on training data only -- transform test data using training parameters.
How the result changes with Maximum Value
| Maximum Value | Min-Max Scaled Value | Z-Score (Standard Scaled) |
|---|---|---|
| 50 | 1 | 0 |
| 75 | 0.67 | 0 |
| 150 | 0.33 | 0 |
| 250 | 0.2 | 0 |
What each input means
- Raw Value
- The original unscaled feature value you want to transform.
- Minimum Value
- The minimum observed value in your feature column.
- Maximum Value
- The maximum observed value in your feature column.
- Mean
- The arithmetic mean (average) of your feature column.
- Standard Deviation
- The standard deviation of your feature column. Used for z-score standardization.
What each result means
- Min-Max Scaled Value
- Value scaled to [0, 1] range using (x - min) / (max - min).
- Z-Score (Standard Scaled)
- Number of standard deviations from the mean: (x - mean) / std.
- Robust Scaled Value
- Scaled using range instead of std dev, less sensitive to outliers.
- Max-Abs Normalized
- Value divided by the maximum absolute value in the range.
How this is calculated
Worked example, using the default values
- Identify Input Parameters4 parametersRaw Value = 50, Minimum Value = 0, Maximum Value = 100, Mean = 50 = 5 input(s) provided
- Calculate Min-Max Scaled ValueMin-Max Scaled Value0.5 = 0.5
- Calculate Z-ScoreZ-Score0 = 0
- Calculate Robust Scaled ValueRobust Scaled Value0 = 0
- Calculate Max-Abs NormalizedMax-Abs Normalized0.5 = 0.5
Engine last updated . Checked against 1 independently-derived test — how we verify calculators. Built by Paul Gunder, a software engineer, not a licensed financial, medical, or legal professional.
Frequently Asked Questions
Why does Deviations from Mean show zero sensitivity to Raw Value, Mean, and Standard Deviation all at once?
Because Raw Value and Mean both default to 50, making the underlying z-score exactly 0 before any scaling happens — and dividing 0 by any Standard Deviation still gives 0. Nudging Raw Value alone up or down by 10% moves the z-score to the same magnitude on either side of 0, and since Deviations from Mean takes the absolute value of that z-score (line 23), the output reads unchanged either way. Change Raw Value away from 50 and Deviations from Mean responds to all three inputs immediately.
Is "Robust Scaled Value" the same as the standard robust scaler used in scikit-learn?
No — the standard robust scaler centers on the median and divides by the interquartile range specifically to resist outliers. This calculator's version (line 13) centers on the Mean and divides by the Minimum-to-Maximum range instead, which is a simpler approximation that a single extreme value in Minimum Value or Maximum Value can still distort, unlike a true IQR-based robust scaler.
Does Standard Deviation affect Min-Max Scaled Value or Robust Scaled Value?
No — neither formula references it. Min-Max Scaled Value is `(rawValue - minVal) / range` (line 11) and Robust Scaled Value is `(rawValue - mean) / range` (line 13); Standard Deviation only enters the calculation for Z-Score (line 12), where it's the divisor.
Does Mean affect Max-Abs Normalized?
No — Max-Abs Normalized divides Raw Value by the larger of |Minimum Value| and |Maximum Value| (lines 14-15), with no reference to Mean or Standard Deviation anywhere in that formula; only Raw Value, Minimum Value, and Maximum Value move this output.
Related Calculators
The questions that sit next to this one — chosen by subject, including calculators filed under a different category.
Outlier Detection Calculator
Detect outliers using IQR bounds, z-score bounds, and modified z-score methods. Determine whether a data point is an outlier using configurable thresholds.
Data ScienceConfusion Matrix Analyzer
Analyze classification model performance from confusion matrix values. Calculate precision, recall, F1 score, MCC, accuracy, specificity, and more.
Data ScienceData Imputation Calculator
Estimate missing data values using mean, median, mode, or linear interpolation imputation methods. Compare imputation accuracy against actual values.
Machine Learning & AIInformation Gain Calculator
Calculate information gain, gain ratio, Gini gain, and feature importance for decision trees.
More in Math & Statistics.