Return Type

The return type of get_samples is determined by the return type of the bfunc function.

bfunc return type	get_samples return type	index of get_samples
float or int	pd.Series	sample number
pd.Series	pd.DataFrame	sample number
pd.DataFrame	pd.DataFrame	sample number and original index

Example

For some concrete examples, we will use the following data:

import pandas as pd
import numpy as np

import bootstrap

n_points = 100
rng = np.random.default_rng(0)
df = pd.DataFrame({
    "a": rng.normal(size=n_points),
    "b": rng.normal(size=n_points),
    "group": rng.choice(["A", "B"], size=n_points)
})

float or int return a pd.Series

def float_return(df: pd.DataFrame, col: str) -> float:
    return df[col].mean()

series = df.boot.get_samples(bfunc=float_return, col="a", B=5)

sample
0    0.111948
1    0.157558
2    0.177292
3    0.000501
4    0.082680
Name: stat, dtype: float64

pd.Series returns a pd.DataFrame

def series_return(df: pd.DataFrame) -> pd.Series:
    return df.mean(numeric_only=True)

dataframe = df.boot.get_samples(bfunc=series_return, B=5)

               a         b
sample
0       0.123708 -0.070613
1      -0.011843  0.030508
2       0.027869 -0.006668
3       0.065575  0.043360
4       0.089570  0.125474

pd.DataFrame returns a pd.DataFrame with another index

def df_return(df: pd.DataFrame) -> pd.DataFrame:
    return df.groupby("group").mean()

dataframe_grouped = df.boot.get_samples(bfunc=df_return, B=5)

                     a         b
group sample
A     0       0.066174 -0.125440
      1      -0.085131 -0.079708
      2       0.128932  0.096542
      3       0.088746 -0.295118
      4      -0.042594  0.063312
B     0      -0.032779  0.035043
      1       0.065471 -0.072222
      2      -0.135504  0.040352
      3       0.120614  0.202248
      4       0.057131  0.027761

Comments