📈↔️Correlation Coefficient

NEW

A numerical measure of the strength and direction of association between two variables, often focusing on linear relationships.

DESCRIPTION

Correlation coefficients quantify how two variables vary together, with values typically between −1 and 1. Pearson’s r measures linear correlation; Spearman’s rho and Kendall’s tau assess monotonic relationships and handle ranks. Correlation does not imply causation and can be distorted by outliers, nonlinearity, or confounding. Proper use includes exploratory analysis, feature selection, and residual diagnostics. Always inspect scatter plots and context before drawing conclusions.

ALTERNATIVE DEFINITIONS

Wikipedia

In statistics, a correlation coefficient is a numerical measure of some type of correlation, meaning a statistical relationship between two variables.

Broad framing covering different types of correlation.

Source

Encyclopaedia Britannica

Correlation coefficient, a measure of the degree of association between two variables.

Concise emphasis on degree of association.

Source

NIST/SEMATECH e-Handbook of Statistical Methods

The correlation coefficient is a measure of the strength of the linear relationship between two variables.

Highlights focus on linear relationships for Pearson’s r.

Source

RELATED TERMS

Pearson Correlation Spearman Rank Correlation Covariance Linear Model Scatter Plot Hypothesis Testing Kendall’s Tau R-squared Causation p-value

COMPETING TERMS

Mutual Information

RELATIONS

• Narrower: Matthews Correlation Coefficient Pearson Correlation Coefficient Spearman’s Rank Correlation Coefficient

KNOWLEDGE AREAS

Mathematics Data Science

TARGET AUDIENCE

Data Scientist Data Analyst BI Specialist

EXTERNAL SOURCES

• Wikipedia - Correlation coefficient
• NIST - Handbook entry on correlation
• Penn State - Course notes on correlation and r
• Khan Academy - Correlation tutorials and videos

TOP QUESTIONS

What does a correlation coefficient of 0 mean?

It indicates no linear association, though nonlinear relationships may still exist.

Category: Definition

How do Pearson and Spearman correlations differ?

Pearson measures linear correlation; Spearman assesses monotonic relationships using rank-transformed data.

Category: Comparison

Why doesn’t correlation imply causation?

Correlation can arise from confounding variables, reverse causality, or coincidence; causal inference needs controlled designs or robust methods.

Category: Misunderstanding

How do outliers affect correlation?

Outliers can inflate or deflate coefficients; use robust methods or inspect scatter plots.

Category: Risk

When should you use Kendall’s tau?

With small samples or many ties, Kendall’s tau can be more reliable than Spearman’s rho.

Category: Guidance

What is a good way to visualize correlation?

Use scatter plots with trend lines, correlation heatmaps, and pair plots for multivariate data.

Category: HowTo

How to test significance of a correlation coefficient?

Use t-tests for Pearson’s r under normality assumptions, or permutation tests for robustness.

Category: Testing

Can correlation be used for feature selection?

Yes, to detect redundancy and multicollinearity, but consider model-specific importance and nonlinearity.

Category: Application

What is autocorrelation versus correlation?

Autocorrelation is correlation of a series with itself at different lags; it reveals temporal dependence.

Category: Definition