Statistical Bootstrap with Pandas made easy.
pip install pandas-bootstrap
The module is very easy to use.
import bootstrap
- define statistic function:
def some_func(df: pd.DataFrame | pd.Series):
- get bootstrapped samples:
df.boot.get_samples(bfunc=some_func, B=100)
Below is a simple example of bootstrapping the mean of two columns.
import pandas as pd
import bootstrap
df = pd.DataFrame({
'a': [1, 2, 3, 4, 5],
'b': [6, 7, 8, 9, 10],
})
def mean_of_columns(df):
return df.mean(numeric_only=True)
sample_kwargs = dict(random_state=42)
df_bootstrap = df.boot.get_samples(bfunc=mean_of_columns, B=5, sample_kwargs=sample_kwargs)
which results in:
a b
sample
0 3.0 8.0
1 2.6 7.6
2 4.0 9.0
3 3.2 8.2
4 3.0 8.0
Read more in the documentation