How to Use corr in pandas: Calculate Correlation Easily

Use the corr() method on a pandas DataFrame to calculate the correlation matrix between its numeric columns. It returns a new DataFrame showing correlation coefficients, which measure how strongly columns relate to each other.

Syntax

The basic syntax of corr() in pandas is:

• DataFrame.corr(method='pearson', min_periods=1)

method: The correlation method to use. Default is 'pearson'. Other options include 'kendall' and 'spearman'.

min_periods: Minimum number of observations required per pair of columns to have a valid result.

python

DataFrame.corr(method='pearson', min_periods=1)

Example

This example shows how to calculate the correlation matrix of a DataFrame with numeric columns using corr(). It helps understand relationships between columns.

python

import pandas as pd

data = {
    'age': [25, 32, 47, 51, 62],
    'income': [50000, 60000, 80000, 90000, 120000],
    'score': [200, 220, 250, 270, 300]
}
df = pd.DataFrame(data)

correlation_matrix = df.corr()
print(correlation_matrix)

Output

             age    income     score
age     1.000000  0.981981  0.981981
income  0.981981  1.000000  1.000000
score   0.981981  1.000000  1.000000

Common Pitfalls

Common mistakes when using corr() include:

• Trying to calculate correlation on non-numeric columns, which will be ignored or cause errors.
• Not handling missing values, which can affect results.
• Assuming correlation implies causation; correlation only shows association strength.

python

import pandas as pd

data = {
    'age': [25, 32, None, 51, 62],
    'income': [50000, 60000, 80000, None, 120000],
    'name': ['Alice', 'Bob', 'Charlie', 'David', 'Eva']
}
df = pd.DataFrame(data)

# corr() automatically ignores non-numeric columns
correlation = df.corr()
print(correlation)

# Missing values are ignored pairwise by default

Output

             age    income
age     1.000000  0.981981
income  0.981981  1.000000

Quick Reference

Summary tips for using corr():

• Use method='pearson' for linear correlation (default).
• Use method='kendall' or method='spearman' for rank-based correlation.
• Missing values are ignored pairwise by default.
• Only numeric columns are considered.

Parameter	Description	Default
method	Correlation method: 'pearson', 'kendall', or 'spearman'	'pearson'
min_periods	Minimum observations required per pair	1
numeric_only	Consider only numeric columns (automatic)	True

Key Takeaways

Use DataFrame.corr() to get correlation matrix of numeric columns easily.

Default method is 'pearson' for linear correlation; others include 'kendall' and 'spearman'.

Non-numeric columns are ignored automatically by corr().

Missing values are handled pairwise and do not cause errors.

Correlation shows association strength, not cause-effect relationships.