How to compute Z-score

scale()

Learn how to compute z-score with this comprehensive R tutorial. Includes practical examples and code snippets.

Published

September 30, 2024

Introduction

A Z-score measures how many standard deviations a data point is from the mean, allowing you to standardize values across different scales. This is essential for comparing data points from different distributions, identifying outliers, and preparing data for statistical analyses that assume standardized inputs.

Getting Started

library(tidyverse)
library(palmerpenguins)

Example 1: Basic Usage

The Problem

We need to calculate Z-scores for penguin body mass to understand which penguins are unusually heavy or light. This will help us identify outliers and standardize the measurements.

Step 1: Prepare the data

First, let’s examine our dataset and remove any missing values.

# Load and inspect penguin data
data("penguins")
clean_penguins <- penguins |> 
  filter(!is.na(body_mass_g)) |> 
  select(species, body_mass_g)

head(clean_penguins)

We now have clean data with 342 penguins and their body masses ready for Z-score calculation.

Step 2: Calculate Z-scores manually

We’ll compute Z-scores using the formula: (value - mean) / standard deviation.

# Calculate mean and standard deviation
mass_mean <- mean(clean_penguins$body_mass_g)
mass_sd <- sd(clean_penguins$body_mass_g)

# Compute Z-scores
clean_penguins$z_score <- (clean_penguins$body_mass_g - mass_mean) / mass_sd

Each penguin now has a Z-score showing how many standard deviations their mass is from the average.

Step 3: Interpret the results

Let’s examine the Z-scores to understand what they tell us.

# View penguins with extreme Z-scores
extreme_penguins <- clean_penguins |> 
  filter(abs(z_score) > 2) |> 
  arrange(desc(z_score))

print(extreme_penguins)

Penguins with Z-scores above 2 or below -2 are unusually heavy or light, representing less than 5% of the population.

Example 2: Practical Application

The Problem

We want to compare body mass across different penguin species by standardizing within each group. This allows us to identify which individual penguins are outliers within their own species, rather than compared to all penguins combined.

Step 1: Group-wise Z-score calculation

We’ll calculate Z-scores separately for each species using group operations.

# Calculate Z-scores within each species
penguins_grouped <- clean_penguins |> 
  group_by(species) |> 
  mutate(
    species_mean = mean(body_mass_g),
    species_sd = sd(body_mass_g)
  )

Now each penguin has their species-specific mean and standard deviation for comparison.

Step 2: Compute species-standardized Z-scores

We’ll create Z-scores that show how unusual each penguin is within its own species.

# Calculate within-species Z-scores
penguins_grouped <- penguins_grouped |> 
  mutate(
    species_z_score = (body_mass_g - species_mean) / species_sd
  ) |> 
  ungroup()

These species-specific Z-scores reveal which penguins are outliers relative to their own kind.

Step 3: Compare overall vs species-specific outliers

Let’s identify penguins that are outliers in different contexts.

# Find different types of outliers
outlier_comparison <- penguins_grouped |> 
  mutate(
    overall_outlier = abs(z_score) > 2,
    species_outlier = abs(species_z_score) > 2
  ) |> 
  select(species, body_mass_g, z_score, species_z_score, 
         overall_outlier, species_outlier)

This comparison reveals how standardization context affects outlier identification.

Step 4: Visualize the differences

Create a visualization to show both standardization approaches.

# Create comparison plot
ggplot(penguins_grouped, aes(x = species, y = species_z_score, fill = species)) +
  geom_boxplot(alpha = 0.7) +
  geom_hline(yintercept = c(-2, 2), linetype = "dashed", color = "red") +
  labs(title = "Species-Standardized Z-scores",
       x = "Species",
       y = "Z-score within species")

Boxplot of species-standardized z-scores of penguin body mass for Adelie, Chinstrap, and Gentoo, with dashed red lines at plus and minus two marking the outlier threshold used in within-group z-score analysis in R.

The plot shows that each species now has a similar distribution of Z-scores, making cross-species comparisons more meaningful.

Summary

Z-scores standardize data by measuring distance from the mean in standard deviation units
Calculate Z-scores using the formula: (value - mean) / standard deviation
Values with absolute Z-scores greater than 2 are typically considered outliers
Group-wise Z-scores allow meaningful comparisons within subgroups of your data
Choose between overall or group-specific standardization based on your analysis goals

--- title: "How to compute Z-score" description: "Learn how to compute z-score with this comprehensive R tutorial. Includes practical examples and code snippets." date: 2024-09-30 categories: ['scale()'] format: html: code-fold: false code-tools: true --- ## Introduction A Z-score measures how many standard deviations a data point is from the mean, allowing you to standardize values across different scales. This is essential for comparing data points from different distributions, identifying outliers, and preparing data for statistical analyses that assume standardized inputs. ## Getting Started ```r library(tidyverse) library(palmerpenguins) ``` ## Example 1: Basic Usage ### The Problem We need to calculate Z-scores for penguin body mass to understand which penguins are unusually heavy or light. This will help us identify outliers and standardize the measurements. ### Step 1: Prepare the data First, let's examine our dataset and remove any missing values. ```r # Load and inspect penguin data data("penguins") clean_penguins <- penguins |> filter(!is.na(body_mass_g)) |> select(species, body_mass_g) head(clean_penguins) ``` We now have clean data with 342 penguins and their body masses ready for Z-score calculation. ### Step 2: Calculate Z-scores manually We'll compute Z-scores using the formula: (value - mean) / standard deviation. ```r # Calculate mean and standard deviation mass_mean <- mean(clean_penguins$body_mass_g) mass_sd <- sd(clean_penguins$body_mass_g) # Compute Z-scores clean_penguins$z_score <- (clean_penguins$body_mass_g - mass_mean) / mass_sd ``` Each penguin now has a Z-score showing how many standard deviations their mass is from the average. ### Step 3: Interpret the results Let's examine the Z-scores to understand what they tell us. ```r # View penguins with extreme Z-scores extreme_penguins <- clean_penguins |> filter(abs(z_score) > 2) |> arrange(desc(z_score)) print(extreme_penguins) ``` Penguins with Z-scores above 2 or below -2 are unusually heavy or light, representing less than 5% of the population. ## Example 2: Practical Application ### The Problem We want to compare body mass across different penguin species by standardizing within each group. This allows us to identify which individual penguins are outliers within their own species, rather than compared to all penguins combined. ### Step 1: Group-wise Z-score calculation We'll calculate Z-scores separately for each species using group operations. ```r # Calculate Z-scores within each species penguins_grouped <- clean_penguins |> group_by(species) |> mutate( species_mean = mean(body_mass_g), species_sd = sd(body_mass_g) ) ``` Now each penguin has their species-specific mean and standard deviation for comparison. ### Step 2: Compute species-standardized Z-scores We'll create Z-scores that show how unusual each penguin is within its own species. ```r # Calculate within-species Z-scores penguins_grouped <- penguins_grouped |> mutate( species_z_score = (body_mass_g - species_mean) / species_sd ) |> ungroup() ``` These species-specific Z-scores reveal which penguins are outliers relative to their own kind. ### Step 3: Compare overall vs species-specific outliers Let's identify penguins that are outliers in different contexts. ```r # Find different types of outliers outlier_comparison <- penguins_grouped |> mutate( overall_outlier = abs(z_score) > 2, species_outlier = abs(species_z_score) > 2 ) |> select(species, body_mass_g, z_score, species_z_score, overall_outlier, species_outlier) ``` This comparison reveals how standardization context affects outlier identification. ### Step 4: Visualize the differences Create a visualization to show both standardization approaches. ```r # Create comparison plot ggplot(penguins_grouped, aes(x = species, y = species_z_score, fill = species)) + geom_boxplot(alpha = 0.7) + geom_hline(yintercept = c(-2, 2), linetype = "dashed", color = "red") + labs(title = "Species-Standardized Z-scores", x = "Species", y = "Z-score within species") ``` ![Boxplot of species-standardized z-scores of penguin body mass for Adelie, Chinstrap, and Gentoo, with dashed red lines at plus and minus two marking the outlier threshold used in within-group z-score analysis in R.](/images/statistics/compute-z-score-in-r-species-boxplot-ggplot.png) The plot shows that each species now has a similar distribution of Z-scores, making cross-species comparisons more meaningful. ## Summary - Z-scores standardize data by measuring distance from the mean in standard deviation units - Calculate Z-scores using the formula: (value - mean) / standard deviation - Values with absolute Z-scores greater than 2 are typically considered outliers - Group-wise Z-scores allow meaningful comparisons within subgroups of your data - Choose between overall or group-specific standardization based on your analysis goals --- ## Related Posts - [How to Compute Z-Score of Multiple Columns](/statistics/compute-z-score-of-multiple-columns.html) - [How to z-score normalization in R](/statistics/how-to-z-score-normalization-in-r.html) - [How to Compute Pearson Correlation of Multiple Variables](/statistics/compute-pearson-correlation-of-multiple-variables.html) - [Compute rowwise mean and standard deviation](/dplyr/compute-rowwise-mean-and-standard-deviation.html) - [dplyr across(): Compute column-wise mean](/dplyr/dplyr-across-compute-column-wise-mean.html)