一文带您了解伪对数(Pseudo-Log)：可视化倾斜数据的黄金方法

import numpy as np
import matplotlib.pyplot as plt


# 伪对数10变换函数
def pseudo_log10(x):
    return np.log(x/2+np.sqrt(x*x/4+1))/np.log(10)


# 定义数据
x = np.linspace(-15, 15, 400)
y_pseudo_log10 = pseudo_log10(x)
y_log10 = np.log10(x[200:])  


# 创建图形
plt.figure(figsize=(10, 6))


# 绘制伪对数10变换的曲线，绘制经典对数10变换的曲线
plt.plot(x, y_pseudo_log10, label='Pseudo-log10 Transformation', color='blue')
plt.plot(x[200:], y_log10, label='Classic log10 Transformation', color='black')


# 添加图例
plt.legend()


# 添加标题和标签
plt.title('Comparison of Pseudo-log10 and Classic log10 Transformations')
plt.xlabel('x')
plt.ylabel('Transformed value')


# 显示图形
plt.grid(True)
plt.show()
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.

import numpy as np
import matplotlib.pyplot as plt


# 生成1000个来自右偏分布的样本
data = np.random.exponential(scale=2, size=1000)


# 定义伪对数变换函数
def pseudo_log(x):
    return np.log(x/2+np.sqrt(x*x/4+1))/np.log(10)


# 对数据应用变换
log_data = np.log(data)
pseudo_log_data = pseudo_log(data)


# 绘制原始数据、对数变换数据和伪对数变换数据
plt.figure(figsize=(12, 8))


# 绘制原始数据
plt.subplot(3, 1, 1)
plt.hist(data, bins=50, color='blue', alpha=0.7, label='Original Data')
plt.legend()
plt.title('Original Data')


# 绘制对数变换数据
plt.subplot(3, 1, 2)
plt.hist(log_data, bins=50, color='green', alpha=0.7, label='Log-Transformed Data')
plt.legend()
plt.title('Log-Transformed Data')


# 绘制伪对数变换数据
plt.subplot(3, 1, 3)
plt.hist(pseudo_log_data, bins=50, color='red', alpha=0.7, label='Pseudo-Log-Transformed Data')
plt.legend()
plt.title('Pseudo-Log-Transformed Data')


plt.tight_layout()
plt.show()
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.

import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns


# Simulate some population density data with a wider range
np.random.seed(42)
data = np.concatenate([np.random.lognormal(mean=0, sigma=2.0, size=(10, 20)), 
                       0-np.random.lognormal(mean=0, sigma=2.0, size=(10, 20))])


# Function to plot the heatmaps
def plot_heatmaps(data):
    fig, axes = plt.subplots(1, 3, figsize=(18, 6))
    
    # Linear color scale
    sns.heatmap(data, ax=axes[0], cmap='viridis', cbar_kws={'label': 'population density / km²'})
    axes[0].set_title('Linear color scale')


    # Log10 transformation
    log_data = np.log10(data + 1)
    sns.heatmap(log_data, ax=axes[1], cmap='viridis', cbar_kws={'label': 'log10(population density / km²)'})
    axes[1].set_title('log10 transformation')
    
    # Pseudo-log transformation (log + sqrt)
    pseudo_log_data = np.log(data + np.sqrt(data**2 + 1))
    sns.heatmap(pseudo_log_data, ax=axes[2], cmap='viridis', cbar_kws={'label': 'pseudo-log(population density / km²)'})
    axes[2].set_title('pseudo-log transformation')
    
    plt.tight_layout()
    plt.show()


plot_heatmaps(data)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.