Course contentsShow
← Data Science
Lesson 1184 of 2,14530. Exploratory Data Analysis (EDA) ProcessFree lesson

Correlation Coefficients in Bivariate Analysis

Measuring linear relationships between numeric variables using Pearson, Spearman, and Kendall methods.

Correlation Coefficients in Bivariate Analysis

You'll learn how to measure the strength and direction of linear relationships between two numeric variables using three key correlation methods.

The Core Idea

When analyzing two numeric variables together, scatter plots show visual patterns, but correlation coefficients give you a numeric measure of how strongly those variables move together. Think of it like grading the relationship on a scale from -1 to +1: values near +1 mean "as one increases, the other increases," values near -1 mean "as one increases, the other decreases," and values near 0 mean "no linear relationship."

Three Main Methods

Pearson correlation is your go-to for linear relationships between normally distributed variables. It measures how well data points fit a straight line. Use this when both variables are continuous and roughly bell-shaped.

Spearman correlation works with ranked data instead of raw values. If your data has outliers or isn't normally distributed, Spearman handles it better because it looks at the order of values rather than their exact measurements. Think of it as asking "do the rankings move together?" rather than "do the values move together?"

Kendall correlation also uses ranks but counts how often pairs of observations agree in their ordering. It's more robust with small samples and ties in the data, though less commonly used than Spearman.

When to Use Which

Use Pearson for clean, normally distributed data. Switch to Spearman or Kendall when you have outliers, skewed distributions, or ordinal (ranked) data.

Key Takeaway: Correlation coefficients quantify linear relationships numerically—Pearson for normal distributions, Spearman and Kendall for ranked or non-normal data—giving you objective measures to complement your scatter plot insights.