How to find numeric columns in Pandas?

To find numeric columns in Pandas, we can use the select_dtypes() method to filter columns based on their data types. This method allows us to specify which numeric types to include or exclude from our DataFrame.

Basic Example

Let's start with a simple example using select_dtypes() ?

import pandas as pd

# Create a DataFrame with mixed data types
df = pd.DataFrame({
    'name': ['John', 'Jacob', 'Tom', 'Tim', 'Ally'],
    'marks': [89, 23, 100, 56, 90],
    'subjects': ["Math", "Physics", "Chemistry", "Biology", "English"]
})

print("Input DataFrame:")
print(df)
print("\nData types:")
print(df.dtypes)

Input DataFrame:
    name  marks    subjects
0   John     89        Math
1  Jacob     23     Physics
2    Tom    100   Chemistry
3    Tim     56     Biology
4   Ally     90     English

Data types:
name        object
marks        int64
subjects     object
dtype: object

Method 1: Using select_dtypes() with Specific Types

Select columns by specifying exact numeric data types ?

import pandas as pd

df = pd.DataFrame({
    'name': ['John', 'Jacob', 'Tom', 'Tim', 'Ally'],
    'marks': [89, 23, 100, 56, 90],
    'subjects': ["Math", "Physics", "Chemistry", "Biology", "English"]
})

# Specify exact numeric types
numeric_types = ['int16', 'int32', 'int64']
numeric_df = df.select_dtypes(include=numeric_types)

print("Numeric columns only:")
print(numeric_df)

Numeric columns only:
   marks
0     89
1     23
2    100
3     56
4     90

Method 2: Using select_dtypes() with 'number'

Use the generic 'number' type to include all numeric columns ?

import pandas as pd
import numpy as np

# Create DataFrame with various numeric types
df = pd.DataFrame({
    'integers': [1, 2, 3, 4],
    'floats': [1.5, 2.7, 3.2, 4.1],
    'names': ['A', 'B', 'C', 'D'],
    'booleans': [True, False, True, False]
})

print("All columns:")
print(df.dtypes)

# Select all numeric columns
numeric_df = df.select_dtypes(include='number')
print("\nNumeric columns:")
print(numeric_df)

All columns:
integers      int64
floats      float64
names        object
booleans       bool
dtype: object

Numeric columns:
   integers  floats
0         1     1.5
1         2     2.7
2         3     3.2
3         4     4.1

Method 3: Getting Column Names Only

If you only need the column names, use the columns attribute ?

import pandas as pd

df = pd.DataFrame({
    'name': ['John', 'Jacob', 'Tom'],
    'age': [25, 30, 28],
    'salary': [50000.0, 60000.0, 55000.0],
    'department': ['IT', 'HR', 'Finance']
})

# Get numeric column names
numeric_columns = df.select_dtypes(include='number').columns.tolist()
print("Numeric column names:", numeric_columns)

# Alternative: using numpy
numeric_cols_np = df.select_dtypes(include=[np.number]).columns.tolist()
print("Using np.number:", numeric_cols_np)

Numeric column names: ['age', 'salary']
Using np.number: ['age', 'salary']

Comparison of Methods

Conclusion

Use select_dtypes(include='number') for finding all numeric columns in a Pandas DataFrame. This method is flexible and handles various numeric data types automatically.