Pandas和裸鼠的不同标准差

Question

我试着为一个数组计算std，我试着用numpy和pandas来寻找std，但是我得到的结果不符合逻辑，我对同一个数组有两个不同的std

为什么会这样

>>> import numpy as np
>>> import pandas as pd

>>> a = np.arange(10)+1
>>> a
array([ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10])
>>> a.std()
2.8722813232690143
>>> b = pd.DataFrame(a)
>>> b.std()
0    3.02765
dtype: float64

Answer 1

区别在于自由度，numpy的默认值是ddof=0，pandas的默认值是ddof=1：

print(a.std())
2.8722813232690143
print(a.std(ddof=0))
2.8722813232690143

print(a.std(ddof=1))
3.0276503540974917

---

b = pd.DataFrame(a)
print(b.std())
0    3.02765
dtype: float64
print(b.std(ddof=1))
0    3.02765
dtype: float64

print(b.std(ddof=0))
0    2.872281
dtype: float64