Regex operations in Pandas

Introduction

Regex helps find patterns in text data. In Pandas, it makes searching and changing text easy and fast.

You want to find rows where a column has emails.

You need to replace phone numbers in a text column with a standard format.

You want to filter data where names start with a certain letter.

You want to split a column by a pattern like commas or spaces.

You want to check if a text column contains a specific pattern.

Syntax

Pandas

df['column'].str.contains('pattern', regex=True)
df['column'].str.replace('pattern', 'new_text', regex=True)
df['column'].str.extract('pattern')
df['column'].str.match('pattern')

Use str accessor to apply regex on text columns.

Set regex=True to treat the pattern as a regex (default in recent pandas versions).

Examples

Find rows where 'email' column contains '@gmail.com'.

Pandas

df['email'].str.contains(r'@gmail\.com')

Remove all non-digit characters from 'phone' column.

Pandas

df['phone'].str.replace(r'\D', '', regex=True)

Extract the first word from the 'name' column.

Pandas

df['name'].str.extract(r'^(\w+)')

Check if 'code' column matches pattern: starts with 'A' followed by exactly 3 digits.

Pandas

df['code'].str.match(r'^A\d{3}$')

Sample Program

This program shows how to use regex in Pandas to filter, clean, extract, and match text data.

Pandas

import pandas as pd

data = {
    'email': ['[e-mail]', '[e-mail]', '[e-mail]', '[e-mail]'],
    'phone': ['[phone]', '987-654-3210', '555 666 7777', '444.333.2222'],
    'name': ['Alice Smith', 'Bob Jones', 'Carol White', 'Dave Black'],
    'code': ['A123', 'B234', 'A999', 'C456']
}

df = pd.DataFrame(data)

# Find emails with gmail
gmail_filter = df['email'].str.contains(r'@gmail\.com')

# Clean phone numbers to digits only
clean_phones = df['phone'].str.replace(r'\D', '', regex=True)

# Extract first name
first_names = df['name'].str.extract(r'^(\w+)')

# Check codes starting with A and 3 digits
code_match = df['code'].str.match(r'^A\d{3}$')

print('Rows with Gmail emails:')
print(df[gmail_filter])

print('\nClean phone numbers:')
print(clean_phones)

print('\nFirst names extracted:')
print(first_names)

print('\nCodes matching pattern:')
print(code_match)

Important Notes

Regex patterns use special symbols like \d for digits and \w for letters.

Remember to double backslashes \\ in Python strings for regex.

Use raw strings r'' to write regex patterns easily.

Summary

Regex in Pandas helps find and change text patterns in columns.

Use str.contains, str.replace, str.extract, and str.match for common tasks.

Always test your regex on sample data to avoid mistakes.

Practice

1. What does the str.contains method do in Pandas when used with a regex pattern?

easy

A. It extracts the first match of the regex pattern from each string.

B. It replaces all matches of the regex pattern with a new string.

C. It sorts the DataFrame rows based on regex matches.

D. It checks if each string in a column contains the regex pattern and returns true or false.

Solution

1. Step 1: Understand str.contains usage

   str.contains checks if the regex pattern appears anywhere in each string of the column and returns a boolean Series.

2. Step 2: Differentiate from other methods

   str.replace changes text, str.extract pulls out matches, and sorting is unrelated.

3. Final Answer:

   It checks if each string in a column contains the regex pattern and returns true or false. -> Option D

4. Quick Check:

   str.contains = Boolean check [OK]

Hint: Contains returns true/false if pattern found in each string [OK]

Common Mistakes:

• Confusing contains with replace or extract
• Thinking contains modifies the data
• Assuming contains returns matched text

2. Which of the following is the correct syntax to replace every sequence of consecutive digits in a Pandas column df['col'] with the character 'X' using regex?

easy

A. df['col'].str.replace('\d+', 'X')

B. df['col'].str.replace('[0-9]', 'X')

C. df['col'].replace('[0-9]', 'X')

D. df['col'].str.extract('[0-9]', 'X')

Solution

1. Step 1: Identify correct method for regex replacement

   Use str.replace on the Series to replace regex matches. df['col'].replace('[0-9]', 'X') is wrong because replace without str is for exact values, not regex.

2. Step 2: Check regex pattern and syntax

   df['col'].str.replace('[0-9]', 'X') matches single digits only ('123' -> 'XXX'). df['col'].str.replace('\d+', 'X') matches sequences of one or more digits ('123' -> 'X').

3. Final Answer:

   df['col'].str.replace('\d+', 'X') -> Option A

4. Quick Check:

   Use str.replace with regex pattern [OK]

Hint: Use str.replace with regex pattern and replacement string [OK]

Common Mistakes:

• Using DataFrame replace instead of Series.str.replace
• Wrong regex pattern syntax
• Using str.extract instead of replace

3. Given the DataFrame:

import pandas as pd
df = pd.DataFrame({'text': ['apple123', 'banana', 'cherry456', 'date789']})
df['numbers'] = df['text'].str.extract('(\d+)')

What will be the content of the new column numbers?

medium

A. ['123', 'banana', '456', '789']

B. ['apple', NaN, 'cherry', 'date']

C. ['123', NaN, '456', '789']

D. [NaN, NaN, NaN, NaN]

Solution

1. Step 1: Understand str.extract with regex

   The regex (\d+) captures one or more digits. str.extract returns the first match or NaN if none.

2. Step 2: Apply regex to each string

   'apple123' matches '123', 'banana' has no digits so NaN, 'cherry456' matches '456', 'date789' matches '789'.

3. Final Answer:

   ['123', NaN, '456', '789'] -> Option C

4. Quick Check:

   Extract digits or NaN if none [OK]

Hint: Extract returns first regex group or NaN if no match [OK]

Common Mistakes:

• Expecting extract to return full string
• Confusing extract with contains or replace
• Not handling NaN for no matches

4. Consider this code snippet:

import pandas as pd
df = pd.DataFrame({'col': ['abc123', 'def456', 'ghi789']})
df['col'].str.match('[a-z]+\d')

What is the error or issue in this code?

medium

A. The regex pattern misses the last digit, so match returns false for all rows.

B. The code raises a SyntaxError due to incorrect regex syntax.

C. The method str.match does not accept regex patterns.

D. The code runs correctly and returns true for all rows.

Solution

1. Step 1: Analyze the regex pattern

   The pattern [a-z]+\d matches one or more lowercase letters followed by exactly one digit from the start of the string.

2. Step 2: Compare pattern with data strings

   The strings 'abc123', 'def456', 'ghi789' have multiple digits at the end, but the pattern only expects one digit. Since str.match matches from the start and the pattern does not cover the full digit sequence, it returns False for all rows.

3. Final Answer:

   The regex pattern misses the last digit, so match returns false for all rows. -> Option A

4. Quick Check:

   str.match matches from start; pattern must cover full prefix for True [OK]

Hint: str.match anchors at start (prefix match), pattern must cover full prefix [OK]

Common Mistakes:

• Thinking str.match requires full string match
• Confusing str.match with str.fullmatch
• Thinking regex syntax is invalid here

5. You have a DataFrame df with a column 'emails' containing email addresses mixed with invalid entries. How would you create a new column 'valid' that is true only for rows where 'emails' contains a valid email pattern using regex in Pandas?

hard

A. df['valid'] = df['emails'].str.extract(r'^[\w\.-]+@[\w\.-]+\.\w+$')

B. df['valid'] = df['emails'].str.contains(r'^[\w\.-]+@[\w\.-]+\.\w+$')

C. df['valid'] = df['emails'].str.replace(r'^[\w\.-]+@[\w\.-]+\.\w+$', 'true')

D. df['valid'] = df['emails'].str.match(r'\w+@\w+')

Solution

1. Step 1: Choose method to check pattern presence

   str.contains returns true/false if regex matches anywhere in string, suitable for validation.

2. Step 2: Use correct regex for full email validation

   The regex ^[\w\.-]+@[\w\.-]+\.\w+$ matches a full email pattern from start to end. df['valid'] = df['emails'].str.extract(r'^[\w\.-]+@[\w\.-]+\.\w+$') extracts matched text, not boolean. df['valid'] = df['emails'].str.replace(r'^[\w\.-]+@[\w\.-]+\.\w+$', 'true') replaces text, not boolean. df['valid'] = df['emails'].str.match(r'\w+@\w+') uses str.match but regex is incomplete and may match partial emails.

3. Final Answer:

   df['valid'] = df['emails'].str.contains(r'^[\w\.-]+@[\w\.-]+\.\w+$') -> Option B

4. Quick Check:

   Use str.contains with full regex for validation [OK]

Hint: Use str.contains with full regex pattern for true/false validation [OK]

Common Mistakes:

• Using str.extract which returns strings, not booleans
• Using str.replace which changes data instead of checking
• Using incomplete regex that matches partial emails