Column-wise running totals
Compute running totals for each metric over ordered rows, such as daily sales or balances.
Pandas · Module
df.cumsum
Computes cumulative sums along a DataFrame axis, returning an equally sized result. By default it accumulates down columns, skips missing values, and can optionally restrict the operation to numeric columns.
How it works
Call DataFrame.cumsum to replace each value with the running total along a selected DataFrame axis. By default it accumulates down columns, skips missing values while retaining the missing position, and can optionally restrict calculation to numeric columns.
Syntax
DataFrame.cumsum(axis=0, skipna=True, numeric_only=False, *args, **kwargs)Parameters
| Name | Type | Description |
|---|---|---|
axis | {0 or ‘index’, 1 or ‘columns’}, default 0 | The index or the name of the axis; 0, None, or 'index' accumulates down columns, while 1 or 'columns' accumulates across rows. Default: 0. |
skipna | bool, default True | Whether to exclude NA/null values. Default: True. |
numeric_only | bool, default False | Whether to include only float, integer, and Boolean columns. Default: False. |
*args | tuple | Additional arguments accepted for compatibility. |
**kwargs | dict | Additional keyword arguments accepted for compatibility. |
Returns
DataFrame or Series of the same size containing the cumulative sum.
Key points
- The method computes a running sum along one DataFrame axis.
- For a DataFrame, the result preserves the original dimensions.
- Axis 0,
None, and'index'select accumulation down the index direction. - Use axis 1, or
'columns', to accumulate horizontally within rows. - The default
skipna=Trueomits missing values from the running total. numeric_only=Trueselects float, integer, and Boolean columns.
Common mistakes
- A completely missing row or column remains missing even when
skipna=True. - With
numeric_only=True, non-numeric columns are excluded from the calculation.
Example: Cumulative sums by column or row
import pandas as pd
df = pd.DataFrame({"sales": [2, 3, 1], "returns": [1, None, 0]})
print(df.cumsum())
print(df.cumsum(axis=1))The first call accumulates values down each column; the second accumulates across each row.
When to use it
Row-wise accumulation
Build per-record subtotals across several numeric fields.
Numeric-column filtering
Process mixed-type tables while excluding text and other non-numeric columns.
Interview questions
QWhat does DataFrame.cumsum() return, and what is its default direction?
It returns a DataFrame of the same shape containing running sums. The default axis is 0, which accumulates down each column.
Follow-up: How would you accumulate across each row? Use axis=1, also expressed as axis='columns'.
QHow does skipna affect missing data?
skipna=True excludes missing values from the accumulation. A missing entry is therefore not added to the running total, while an entirely missing row or column produces missing output.
Follow-up: What changes with skipna=False? Missing values participate in propagation, so subsequent cumulative results can become missing after a missing value.
QWhat does numeric_only control?
numeric_only=True limits the operation to float, integer, and Boolean columns. The default is False, so the method does not request numeric-column filtering.
Follow-up: When is this useful? It prevents non-numeric columns from being included when a DataFrame contains mixed data types.
Related
DataFrame.sumSeries.cumsumDataFrameGroupBy.cumsum