Pandas Correlation of Columns

pandas.DataFrame.corr() function can be used to get the correlation between two or more columns in DataFrame. Correlation is used to analyze the strength and direction between two quantitative variables. It is denoted by r and values between -1 and +1. A positive value for r indicates a positive association and a negative value for r indicates a negative association

In this article, I will explain how to get the correlation between two columns with several examples.

Key Points –

• The corr() method is used to calculate the Pearson correlation coefficient between numerical columns in a DataFrame, which measures the linear relationship between columns.
• Calling df.corr() on a DataFrame returns a correlation matrix that shows the pairwise correlation between all numeric columns.
• By default, corr() calculates the Pearson correlation, which ranges from -1 (perfect negative correlation) to 1 (perfect positive correlation), with 0 indicating no linear correlation.
• Pandas corr() also supports other correlation methods, such as Kendall and Spearman, which can be specified using the method parameter.
• You can compute the correlation between specific columns by selecting them before applying corr(), or by indexing into the correlation matrix.
• The corr() method only works with numeric data, meaning it will ignore non-numeric columns during the correlation calculation.

Quick Examples of Correlation of Columns

If you are in hurry below are some quick examples of pandas correlation between two columns.

# Quick examples of correlation of columns

# Example 1: Correlation between two columns of DataFrame
corr=df['Fee'].corr(df['Discount'])

# Example 2: Correlation between all the columns of DataFrame
df2=df.corr()

# Example 3: Other example
df['Discount']=np.float64(df['Fee'])

Now, let’s create a pandas DataFrame with a few rows and columns, execute these examples and validate results. Our DataFrame contains column names Courses, Fee and Duration.

# Create a pandas DataFrame
import pandas as pd
import numpy as np
technologies= {
    'Courses':["Spark","PySpark","Spark","Python","PySpark"],
    'Fee' :[22000,25000,23000,24000,26000],
    'Discount':[1500,1000,1200,800,1300],
    'Duration':['30days','50days','30days','35days','40days']
          }
df = pd.DataFrame(technologies)
print("Create DataFrame:\n",df)

Yields below output.

DataFrame corr() correlation Syntax

Following is the syntax of the DataFrame.corr() function.

# Syntax of corre() 
DataFrame.corr(method='pearson', min_periods=1)

Correlation Between Two Columns of DataFrame

You can see the correlation between two columns of pandas DataFrame by using DataFrame.corr() function. The pandas.DataFrame.corr() is used to find the pairwise correlation of all columns in the DataFrame. For example, let’s see what is the correlation between Fee and Discount.

# Correlation between two columns of DataFrame
corr=df['Fee'].corr(df['Discount'])
print("Correlation between two columns:\n",corr)

Yields below output.

We get -0.35 as the correlation between the scores of Fee and Discount. This indicates that the two columns highly correlated in a negative direction.

Correlation Between All the Columns of DataFrame

You can also get the correlation between all the columns of a pandas DataFrame. For this, apply corr() function on the entire DataFrame which will result in a DataFrame of pair-wise correlation values between all the columns.

Note that by default, the corr() function returns Pearson’s correlation.

# Correlation between all the columns of DataFrame.
df2=df.corr()
print(df2)

Yields below output.

# Output:
               Fee  Discount
Fee       1.000000 -0.351123
Discount -0.351123  1.000000

When applied to an entire DataFrame, the corr() function returns a DataFrame of pair-wise correlation between the columns. We can see that there’s a weak negative correlation between scores of Fee/Discount. Also, notice that the values on the diagonal are 1s, this is because each column is perfectly correlated with itself.

Other Example

In this example, if Fee is float type, python skips it by default. All the other columns of DataFrame are in numpy-formats. so, you can do it by converting the column to np.float64.

# Other example
df['Fee']=np.float64(df['Fee'])
print(df)

Complete Examples of Correlation Between Two Columns

import pandas as pd
import numpy as np
technologies= {
    'Courses':["Spark","PySpark","Spark","Python","PySpark"],
    'Fee' :[22000,25000,23000,24000,26000],
    'Discount':[1500,1000,1200,800,1300],
    'Duration':['30days','50days','30days','35days','40days']
          }
df = pd.DataFrame(technologies)
print(df)

# Correlation between two columns of DataFrame
corr=df['Fee'].corr(df['Discount'])
print(corr)

# Correlation between all the columns of DataFrame
df2=df.corr()
print(df2)

# Other example.
df['Fee']=np.float64(df['Fee'])
df2=df.corr()

Frequently Asked Questions on Pandas Correlation of Columns

Conclusion

In this article, you have learned how to get the correlation between two columns by using DataFrame.corr() method which can get positive and negative values between columns with several examples.

Related Articles

• Split pandas DataFrame
• Pandas Split Column into Two Columns
• How to Convert pandas Column to List
• Pandas Convert Datetime to Date Column
• Split Pandas DataFrame by column value
• Convert Pandas DataFrame to JSON String
• Pandas Find Row Values for Column Maximal
• Get Pandas DataFrame Columns by Data Type
• Pandas Add Column based on Another Column
• Pandas Check Column Contains a Value in DataFrame
• Count(Distinct) SQL Equivalent in Pandas DataFrame
• Create Test and Train Samples from Pandas DataFrame

References

• https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.corr.html