The Bootstrap¶

In [1]:

from sklearn.utils import resample
import pandas as pd

from sklearn.utils import resample import pandas as pd

In [2]:

loans_income = pd.read_csv("../data/loans_income.csv")
loans_income

loans_income = pd.read_csv("../data/loans_income.csv") loans_income

Out[2]:

	x
0	67000
1	52000
2	100000
3	78762
4	37041
...	...
49995	40000
49996	54000
49997	50000
49998	82000
49999	70000

50000 rows × 1 columns

In [3]:

results_bootstrap = []
results= []

sample_from_income = loans_income.sample(150)

for _ in range(1000):
    sample_bootstrap = resample(sample_from_income)
    sample = loans_income.sample(150)
    results.append(sample["x"].mean())
    results_bootstrap.append(sample_bootstrap["x"].mean())

res_df = pd.Series(results_bootstrap)
res_df_real = pd.Series(results)

results_bootstrap = [] results= [] sample_from_income = loans_income.sample(150) for _ in range(1000): sample_bootstrap = resample(sample_from_income) sample = loans_income.sample(150) results.append(sample["x"].mean()) results_bootstrap.append(sample_bootstrap["x"].mean()) res_df = pd.Series(results_bootstrap) res_df_real = pd.Series(results)

In [4]:

print(f"Original mean: {loans_income["x"].mean()}")
print(f"Real sampling mean: {res_df_real.mean()}")
print(f"Bootstrap method mean {res_df.mean()}")
print(f"Bias: {res_df.mean()- loans_income["x"].mean()}")
print(f"Bootstrapped Std. Error: {res_df.std()}")
print(f"Real Std. Error: {res_df_real.std()}")

print(f"Original mean: {loans_income["x"].mean()}") print(f"Real sampling mean: {res_df_real.mean()}") print(f"Bootstrap method mean {res_df.mean()}") print(f"Bias: {res_df.mean()- loans_income["x"].mean()}") print(f"Bootstrapped Std. Error: {res_df.std()}") print(f"Real Std. Error: {res_df_real.std()}")

Original mean: 68760.51844
Real sampling mean: 68829.19609333333
Bootstrap method mean 66695.42792666666
Bias: -2065.0905133333436
Bootstrapped Std. Error: 2412.326225776251
Real Std. Error: 2751.5871297950516

In [ ]: