Formula & Calculator
Z-Score
Standardizes a data point by expressing how many standard deviations it is from the mean, enabling comparison across different scales.
Interpretation
z = (x − μ)/σ. Number of standard deviations a value is from the mean. Used to standardise data, detect outliers, and compare across different distributions.
Variables
| Symbol | Quantity | Unit |
|---|---|---|
| z | Z-score | |
| x | Data value | |
| μ | Population mean | |
| σ | Population standard deviation |
What it means
A z‑score (or standard score) indicates how many standard deviations an observation is above or below the mean. It transforms data to a standard normal distribution (mean 0, standard deviation 1). Z‑scores are used to compare observations from different distributions, to identify outliers (|z| > 3), and to calculate percentiles. They are also the basis for many statistical tests (z‑test). In quality control, z‑scores are used to monitor processes. In education, they are used to standardise test scores. Understanding z‑scores is essential for interpreting data relative to a distribution and for standardising variables in regression.
Worked example
Z‑Score – Two Examples
Real‑World| Parameter | Value |
|---|---|
| x | 85 |
| μ | 75 |
| σ | 10 |
| Parameter | Value |
|---|---|
| x | 60 |
| μ | 75 |
| σ | 10 |
Common mistakes
- Z‑score: Measures how many standard deviations a value is from the mean.
- Mean μ and standard deviation σ: Use population parameters – for sample, use x̄ and s.
- Sign: Positive means above the mean; negative means below.
- Units: The z‑score is dimensionless.
- Distribution: The z‑score is used for standard normal distribution – but can be applied to any data.
Applications
The Z‑score expresses a data point as the number of standard deviations from the mean, standardising values across different scales. It is used to compare observations from different populations, to detect outliers, and to compute probabilities in the normal distribution. Engineers use Z‑scores to set tolerance limits, to monitor process performance, and to standardise measurements. In finance, it assesses credit risk (Altman Z‑score). In education, it compares test scores. By using Z‑scores, professionals can determine how unusual a value is, identify anomalies, and make cross‑dataset comparisons. This transformation is fundamental to many statistical methods, including standardisation in machine learning.
- Outlier detection and anomaly identification
- Standardisation of variables for multivariate analysis
- Process capability indices (Cpk calculation)
- Credit scoring and risk assessment (Altman Z‑score)
- Educational testing – comparing scores across different tests
Frequently Asked Questions
z = (x − μ) / σ. It standardizes a data point by expressing how many standard deviations it is away from the mean. A positive z means above average; negative means below.
It allows comparison of values from different distributions or different units. It is also used to identify outliers and to compute probabilities under the normal distribution.
A z‑score of 1.5 means the data point is 1.5 standard deviations above the mean. In a normal distribution, this corresponds to about the 93rd percentile.
Z‑scores can be converted to percentiles using the standard normal CDF. For example, z = 1.96 corresponds to the 97.5th percentile (two‑tailed).
In a z‑test, the test statistic is a z‑score that measures how many standard errors the sample mean is from the hypothesized mean. It is used to compute p‑values.
If the original data is normally distributed, the z‑scores have a standard normal distribution with mean 0 and standard deviation 1. The transformation is linear.
They can still be computed, but the interpretation in terms of percentiles requires knowledge of the distribution. They are often used as a scaling technique regardless of normality.
A t‑score is used when the population standard deviation is unknown and is estimated from the sample. It follows a t‑distribution, which has heavier tails for small samples.
For the sampling distribution of the mean, the z‑score is z = (x̄ − μ) / (σ/√n), which is the standard error of the mean.
Z‑score normalization (standardization) is a common preprocessing step to bring features to a similar scale, which improves the performance of many algorithms like SVM, PCA, and neural networks.