Combining and reshaping Updated November 2026

pivot

df.pivot(index="id", columns="key", values="value") reshapes long data into wide: one row per index value and one column per distinct key. Each index/key pair must appear once; if there are duplicates, use pivot_table, which aggregates them.

Syntax

df.pivot(index=None, columns=None, values=None)

Part	What it is
`index`	Column whose values become the row labels
`columns`	Column whose distinct values become the new column names
`values`	Column(s) that fill the cells; omit to spread all remaining columns

Examples

import pandas as pd

long = pd.DataFrame({"date": ["2024-01", "2024-01", "2024-02", "2024-02"],
                     "city": ["Oslo", "Rome", "Oslo", "Rome"],
                     "temp": [-3, 8, -1, 10]})
long.pivot(index="date", columns="city", values="temp")

Result: city Oslo Rome date 2024-01 -3 8 2024-02 -1 10

wide = long.pivot(index="date", columns="city", values="temp")
wide.reset_index().rename_axis(columns=None)

Result: date Oslo Rome 0 2024-01 -3 8 1 2024-02 -1 10

reset_index turns date back into a column; rename_axis(columns=None) removes the city label above the header.

long.set_index(["date", "city"])["temp"].unstack()

Result: city Oslo Rome date 2024-01 -3 8 2024-02 -1 10

unstack on a two-level index does the same job as pivot.

Common mistakes

ValueError: Index contains duplicate entries, cannot reshape: a date/city pair occurs more than once; use pivot_table with an aggfunc, or drop the duplicates first.
The result has a two-level column header: you omitted values with several remaining columns; pass values="temp".
KeyError: you passed positional arguments in the old order; use the keyword names index=, columns=, values=.

Practice pivot

These drills check your answer on live data, free, no account needed.

Long to wide with pivotReshape `long` into a table with one row per store and one column per quarter, holding the sales values.
Index level into columns with unstack`s` holds revenue indexed by region and quarter. Use `unstack` so regions are rows and quarters are columns. Missing combinations stay NaN.

Questions

pivot or pivot_table?

pivot when each row/column pair is unique and you only want to reshape. pivot_table when you need to aggregate (sum, mean) or when duplicates may exist.

How do I pivot without an index column?

Leave index=None and pandas uses the existing index; usually you want to reset_index first so the row labels are meaningful.

pivot_tableHow to build a pivot table in pandas with pivot_table: index, columns, values, aggfunc, fill_value for empty cells, margins for totals, several aggregations.
meltHow to use melt in pandas to turn wide data (one column per month or category) into long format with a variable and a value column, keeping id columns.
set_index and reset_indexHow to use set_index and reset_index in pandas: make a column the row index, go back to a plain column, drop the index, and flatten a groupby result.
groupbyHow groupby works in pandas: group by one or more columns, apply sum, mean, count or size, get a flat table with as_index=False, and iterate over groups.