Python Pandas - Remove the specified categories from CategoricalIndex

To remove the specified categories from CategoricalIndex, use the remove_categories() method in Pandas. This method removes categories from the index and sets values that were in the removed categories to NaN.

Creating a CategoricalIndex

First, let's create a CategoricalIndex with some categories ?

import pandas as pd

# Create CategoricalIndex with categories p, q, r, s
cat_index = pd.CategoricalIndex(
    ["p", "q", "r", "s", "p", "q", "r", "s"], 
    ordered=True, 
    categories=["p", "q", "r", "s"]
)

print("Original CategoricalIndex:")
print(cat_index)
print("\nCategories:")
print(cat_index.categories)

Original CategoricalIndex:
CategoricalIndex(['p', 'q', 'r', 's', 'p', 'q', 'r', 's'], categories=['p', 'q', 'r', 's'], ordered=True, dtype='category')

Categories:
Index(['p', 'q', 'r', 's'], dtype='object')

Removing Categories

Use remove_categories() to remove specific categories. Values that belonged to the removed categories will become NaN ?

import pandas as pd

cat_index = pd.CategoricalIndex(
    ["p", "q", "r", "s", "p", "q", "r", "s"], 
    ordered=True, 
    categories=["p", "q", "r", "s"]
)

# Remove categories 'p' and 'q'
updated_index = cat_index.remove_categories(["p", "q"])

print("CategoricalIndex after removing 'p' and 'q':")
print(updated_index)
print("\nRemaining categories:")
print(updated_index.categories)

CategoricalIndex after removing 'p' and 'q':
CategoricalIndex([nan, nan, 'r', 's', nan, nan, 'r', 's'], categories=['r', 's'], ordered=True, dtype='category')

Remaining categories:
Index(['r', 's'], dtype='object')

Key Points

• The remove_categories() method permanently removes categories from the CategoricalIndex
• Values that were in the removed categories are automatically set to NaN
• The method returns a new CategoricalIndex object with the updated categories
• The order of remaining categories is preserved if the original index was ordered

Conclusion

The remove_categories() method provides a clean way to remove unwanted categories from a CategoricalIndex. Remember that values in the removed categories will become NaN, so handle missing data appropriately in your analysis.