How to reshape a pandas DataFrame from wide to long

Wide tables often place one repeated measure in each column, which makes periods or categories part of the schema instead of row data. Converting those columns into rows gives plotting, grouping, and joins a consistent variable column and a single value column.

Pandas preserves columns named in id_vars and unpivots columns named in value_vars when DataFrame.melt() runs. The selected column labels become values in the var_name column, while their cells become values in the value_name column.

The sales table keeps region and product as identifiers and reshapes three quarterly columns. Assertions check the expected row count, the quarter labels, and the preserved sales total before the script prints the long-form table.

1. Create melt_sales.py with the wide sales table.

   melt_sales.py

   import pandas as pd
    
   wide = pd.DataFrame(
       {
           "region": ["North", "South"],
           "product": ["Widget", "Widget"],
           "q1_sales": [120, 90],
           "q2_sales": [135, 104],
           "q3_sales": [148, 110],
       }
   )

2. Append the identifier and measurement column lists to melt_sales.py.

   id_columns = ["region", "product"]
   value_columns = ["q1_sales", "q2_sales", "q3_sales"]

3. Append the melt operation with names for the generated columns.

   long = wide.melt(
       id_vars=id_columns,
       value_vars=value_columns,
       var_name="quarter",
       value_name="sales_usd",
   )

   value_name must not match an existing column label. Omitting value_vars melts every column not listed in id_vars.

4. Append the outcome checks and long-table display.

   expected_rows = len(wide) * len(value_columns)
   assert len(long) == expected_rows
   assert set(long["quarter"]) == set(value_columns)
   assert long["sales_usd"].sum() == wide[value_columns].sum().sum()
    
   print(long.to_string(index=False))
   print()
   print(f"rows: {len(long)}")
   print(f"columns: {', '.join(long.columns)}")
   print(f"sales total: {long['sales_usd'].sum()}")

5. Run melt_sales.py to verify the reshaped rows and preserved sales total.

   $ python3 melt_sales.py
   region product  quarter  sales_usd
    North  Widget q1_sales        120
    South  Widget q1_sales         90
    North  Widget q2_sales        135
    South  Widget q2_sales        104
    North  Widget q3_sales        148
    South  Widget q3_sales        110
   
   rows: 6
   columns: region, product, quarter, sales_usd
   sales total: 707