Running historical lows
Track the lowest observed value to date separately for each column in ordered records.
Pandas · Module
df.cummin
Returns the cumulative minimum along a DataFrame axis, preserving the input shape. By default it operates down columns, skips missing values, and can optionally restrict processing to numeric and Boolean columns.
How it works
DataFrame.cummin() returns a same-shaped object whose values are the running minimum along the selected axis. With the default index axis, each column is evaluated independently; axis=1 instead evaluates values across each row. Missing values are skipped by default, while numeric_only can restrict processing to numeric and Boolean columns.
Syntax
DataFrame.cummin(axis=0, skipna=True, numeric_only=False, **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' selects the index axis, while 1 or 'columns' selects the columns axis. Default: 0. |
skipna | bool, default True | Whether to exclude missing values. Default: True. |
numeric_only | bool, default False | Whether to include only float, integer, and Boolean columns. Default: False. |
**args, **kwargs | variadic | Additional compatibility arguments accepted by the underlying implementation. |
Returns
DataFrame or Series: cumulative minimum of the input DataFrame or Series.
Key points
- The method calculates a running minimum along one selected DataFrame axis.
- The result preserves the input object's shape rather than reducing it to one aggregate value.
- Axis 0, None, and the index label select the default down-column direction.
- Setting axis=1 changes the operation to a left-to-right running minimum within each row.
Common mistakes
- Skipping missing values does not turn an entirely missing row or column into a numeric result; it remains missing.
- With numeric_only=True, non-numeric columns are excluded from the calculation.
Example: Cumulative minima along both axes
import pandas as pd
df = pd.DataFrame({"A": [2, 3, 1], "B": [1, None, 0]})
print(df.cummin())
print(df.cummin(axis=1))Computes running minima down columns by default, then across each row when axis=1.
When to use it
Row-wise minima
Find progressively smaller values across fields within each record.
Numeric-column filtering
Avoid applying the operation to text or other nonnumeric fields in mixed-type data.
Interview questions
QWhat does DataFrame.cummin() return, and what is its default direction?
It computes the cumulative minimum along a DataFrame axis and returns a DataFrame of the same size. The default axis is 0, so each column receives its running minimum independently.
Follow-up: What changes with axis=1? Each row is processed from left to right, producing the running minimum across that row instead of down columns.
QHow does skipna affect cumulative minima?
skipna=True is the default, so missing values are excluded when possible. However, an entirely missing row or column still produces missing output.
Follow-up: What is a practical consequence? A later non-missing value can resume the cumulative calculation after a missing value when skipna remains enabled.
QWhat does numeric_only control?
numeric_only=True limits the operation to float, integer, and Boolean columns. The default is False, so the operation is not restricted to those column types.
Follow-up: When would you use it? Use numeric_only when a mixed-type DataFrame should be processed only through its numeric and Boolean columns.
Related
DataFrame.cummaxDataFrame.cumprodDataFrame.cumsumSeries.cummin