fit方法估计的是将分布参数化所需的值，而不是关于分布的统计量。实际上，rv_continuous的文档说明：
参数元组：浮点元组。任何形状参数的估计值（如果适用），后跟位置和比例的估计值...
truncnorm分配有两个shape参数：a、b。因此，您收到的元组为：(fitted_a, fitted_b, fitted_loc, fitted_scale)。这些值是首先创建截断正态分布所需的值。truncnorm文档清楚地说明了a和b是什么。下面演示了从查看PDF、从该分布生成随机变量到估计该分布的参数的往返过程。

import numpy as np
from scipy.stats import truncnorm
import matplotlib.pyplot as plt

myclip_a = 0.5
myclip_b = 2.5
loc = 1.3
scale = 1.1
a, b = (myclip_a - loc) / scale, (myclip_b - loc) / scale

x = np.linspace(-2, 4, 100)

# create a graph of how a, b, loc, scale are used to parameterise the truncnorm PDF.

plt.plot(x, truncnorm.pdf(x, a, b, loc, scale),
         'r-', lw=5, alpha=0.6, label='truncnorm pdf')

# create some random variates that obey that truncated normal distribution

noise = truncnorm.rvs(a, b, loc, scale, size=1000000)

# histogram the random variates to visualise the truncated normal distribution.

plt.hist(noise, bins=100);

# now estimate what the values of a, b, loc, scale are from the random variates

fit_a, fit_b, fit_loc, fit_scale = truncnorm.fit(noise, -1,1.2, loc=1, scale=1)

fit_myclip_a = fit_a*fit_scale + fit_loc
fit_myclip_b = fit_b*fit_scale + fit_loc

print(f"Original myclip_a: {myclip_a}, fit_myclip_a: {fit_myclip_a}")
print(f"Original myclip_b: {myclip_b}, fit_myclip_b: {fit_myclip_b}")
print(f"Original loc: {loc}, fit_loc: {fit_loc}")
print(f"Original scale: {scale}, fit_scale: {fit_scale}")

其输出为：

Original myclip_a: 0.5, fit_myclip_a: 0.5000021691860824
Original myclip_b: 2.5, fit_myclip_b: 2.499998152872373
Original loc: 1.3, fit_loc: 1.243679897703037
Original scale: 1.1, fit_scale: 0.9391439098572698