Measure the strength of a monotonic relationship between two variables, even when it's not linear.
Spearman's correlation is Pearson's flexible cousin. Instead of checking for a straight-line relationship, it checks for a **monotonic** one—does one variable consistently increase or decrease as the other does, even if not at a constant rate? It works by converting values to ranks first.
Use Spearman's when the relationship between your variables is not linear, or when your data has significant outliers that would skew a Pearson correlation. It's perfect for capturing relationships that "level off" or accelerate.
A scatter plot can reveal relationships that aren't linear. Spearman's can detect a strong relationship even if the points form a curve.
Example: Let's say we're analyzing the relationship between a custom 'Market Sentiment Score' and a stock's daily return. The return might increase faster as sentiment gets very high. A linear model (Pearson) would miss this, but Spearman's would capture the strong monotonic trend.