How to sort a pandas DataFrame

Row order often determines whether a ranked table, grouped review, or export is easy to interpret. The pandas sort_values() method reorders complete rows from one or more DataFrame columns while keeping every value aligned with its record.

Sort keys are applied from left to right. A matching ascending list can group regions alphabetically and then rank totals from largest to smallest within each region.

Missing values sort last unless na_position=“first” is selected. Because sort_values() returns a new DataFrame by default, assigning the result preserves the source order for a direct comparison.

1. Create the input DataFrame in sort_dataframe.py.

   sort_dataframe.py

   import pandas as pd
    
   orders = pd.DataFrame(
       {
           "order_id": [1042, 1038, 1040, 1041, 1039],
           "region": ["West", "East", "East", "North", "West"],
           "total_usd": [275.0, 410.0, None, 365.0, 190.0],
       }
   )
    
   source_order = orders["order_id"].tolist()

2. Append a descending one-column sort after the DataFrame definition.

   largest_first = orders.sort_values(
       "total_usd",
       ascending=False,
       na_position="last",
   )

   The assigned largest_first DataFrame receives the sorted rows while orders keeps its original order.

3. Append a multi-column sort after the one-column sort.

   sorted_orders = orders.sort_values(
       ["region", "total_usd"],
       ascending=[True, False],
       na_position="last",
       ignore_index=True,
   )

   Each ascending value controls the key in the same position: region sorts A to Z, then total_usd sorts largest to smallest within each region. ignore_index=True replaces the old row labels with a new integer index.

4. Append fail-capable order checks and output after both sort operations.

   assert largest_first["order_id"].tolist() == [1038, 1041, 1042, 1039, 1040]
   assert sorted_orders["order_id"].tolist() == [1038, 1040, 1041, 1042, 1039]
   assert orders["order_id"].tolist() == source_order
    
   print(sorted_orders.to_string(index=False))
   print(f"sorted order IDs: {sorted_orders['order_id'].tolist()}")
   print(f"source order IDs: {orders['order_id'].tolist()}")

5. Confirm the multi-key row order with the completed sorting program.

   $ python3 sort_dataframe.py
    order_id region  total_usd
        1038   East      410.0
        1040   East        NaN
        1041  North      365.0
        1042   West      275.0
        1039   West      190.0
   sorted order IDs: [1038, 1040, 1041, 1042, 1039]
   source order IDs: [1042, 1038, 1040, 1041, 1039]