Example – statistical test

JupyterNotebook (JupyterHub)에서 아래 code를 실행하면서 통계 분석 과정을 이해합니다.

Libraroy import

import numpy as np
import matplotlib.pyplot as plt
from scipy import stats
import seaborn as sns

임의 변수 생성

x = np.random.rand(100)
y = np.random.rand(100)
z = x + (0.5 - np.random.rand(100)) * 0.5

변수 값 시각화

plt.style.use('_mpl-gallery')

fig, ax = plt.subplots()
ax.scatter(x, y, alpha=0.5, label='y')
ax.scatter(x, z, alpha=0.5, label='z')
ax.legend()

Pearson’s correlation test

rho, pv = stats.pearsonr(x, y)
print('y', rho, pv)
rho, pv = stats.pearsonr(x, z)
print('z', rho, pv)

Normality tests

fig = plt.figure(figsize=(7, 3))
ax = fig.add_subplot(1,3,1)
ax2 = fig.add_subplot(1,3,2)
ax3 = fig.add_subplot(1,3,3)
sns.histplot(x, label='x', bins=20, ax=ax)
sns.histplot(y, label='y', bins=20, ax=ax2)
sns.histplot(z, label='y', bins=20, ax=ax3)
fig.tight_layout()

Normality tests – QQ plot

fig = plt.figure(figsize=(7, 3))
ax = fig.add_subplot(1,3,1)
ax2 = fig.add_subplot(1,3,2)
ax3 = fig.add_subplot(1,3,3)
stats.probplot(x, dist=stats.norm, plot=ax)
stats.probplot(y, dist=stats.norm, plot=ax2)
stats.probplot(z, dist=stats.norm, plot=ax3)

Normality tests – Shapiro-Wilk test

# Shapiro-Wilk test 
st, pv = stats.shapiro(x)
print('x', st, pv)
st, pv = stats.shapiro(y)
print('y', st, pv)
st, pv = stats.shapiro(z)
print('z', st, pv)

Spearman’s correlation test (for non-normally distributed variable)

rho, pv = stats.spearmanr(x, y)
print('y', rho, pv)
rho, pv = stats.spearmanr(x, z)
print('z', rho, pv)