Visualize correlation matrices in Python

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

本帖隐藏的内容

When working with data it is helpful to build a correlation matrix to describe data and the associations between variables. In this article, you learn how to use visualizations for correlation matrices in Python.

Read packages into R library

First we need to read the packages into the Python library:

1. 
   
2. # Read packages into Python library
   
3. 
   
4. import pandas as pd
   
5. 
   
6. import numpy as np
   
7. 
   
8. )

复制代码

Visualizations for correlation matrix

First let us make a correlation matrix table:

1. # Create simulated datset in Python:
   
2. 
   
3. rs = np.random.RandomState(0)
   
4. 
   
5. df = pd.DataFrame(rs.rand(10, 10))
   
6. 
   
7. # Create and print correlation matrix:
   
8. 
   
9. corr = df.corr()
   
10. 
    
11. print(corr)
    
12. 
    
13.           0         1         2         3         4         5         6  \
    
14. 
    
15. 0  1.000000  0.347533  0.398948  0.455743  0.072914 -0.233402 -0.731222   
    
16. 
    
17. 1  0.347533  1.000000 -0.284056  0.571003 -0.285483  0.382480 -0.362842   
    
18. 
    
19. 2  0.398948 -0.284056  1.000000 -0.523649  0.152937 -0.139176 -0.092895   
    
20. 
    
21. 3  0.455743  0.571003 -0.523649  1.000000 -0.225343 -0.227577 -0.481548   
    
22. 
    
23. 4  0.072914 -0.285483  0.152937 -0.225343  1.000000 -0.104438 -0.147477   
    
24. 
    
25. 5 -0.233402  0.382480 -0.139176 -0.227577 -0.104438  1.000000 -0.030252   
    
26. 
    
27. 6 -0.731222 -0.362842 -0.092895 -0.481548 -0.147477 -0.030252  1.000000   
    
28. 
    
29. 7  0.477978  0.642578  0.016266  0.473286 -0.523283  0.417640 -0.494440   
    
30. 
    
31. 8 -0.442621  0.252556 -0.434016  0.279258 -0.614603  0.205851  0.381407   
    
32. 
    
33. 9  0.015185  0.190047 -0.383585  0.446650 -0.189916  0.095084 -0.353652   
    
34. 
    
35. 
    
36. 
    
37.           7         8         9  
    
38. 
    
39. 0  0.477978 -0.442621  0.015185  
    
40. 
    
41. 1  0.642578  0.252556  0.190047  
    
42. 
    
43. 2  0.016266 -0.434016 -0.383585  
    
44. 
    
45. 3  0.473286  0.279258  0.446650  
    
46. 
    
47. 4 -0.523283 -0.614603 -0.189916  
    
48. 
    
49. 5  0.417640  0.205851  0.095084  
    
50. 
    
51. 6 -0.494440  0.381407 -0.353652  
    
52. 
    
53. 7  1.000000  0.375873  0.417863  
    
54. 
    
55. 8  0.375873  1.000000  0.150421  
    
56. 
    
57. 9  0.417863  0.150421  1.000000
    

复制代码

The above table is quite hard to read and you end up with a lot of correlation numbers that is hard to interpret. Let us make them into a correlation matrix visualization:

1. # 'RdBu_r' & 'BrBG' are other good diverging colormaps
   
2. 
   
3. corr.style.background_gradient(cmap='coolwarm')
   

复制代码

The above coding gives us the following correlation matrix visualization:

It is possible to limit the digits with the following code:

1. 
   
2. # Limit number of digits:
   
3. 
   
4. corr.style.background_gradient(cmap='coolwarm').set_precision(2)

复制代码

The above coding gives us the following correlation matrix visualization:

It is also possible not to display digits:

1. # Without digits:
   
2. 
   
3. corr.style.background_gradient(cmap='coolwarm').set_properties(**{'font-size': '0pt'})
   

复制代码

The above coding gives us the following correlation matrix visualization:

It is also possible to use another colormap:

1. # Limit number of digits with another colormap:
   
2. 
   
3. corr.style.background_gradient(cmap='viridis').set_precision(2)

复制代码

The above coding gives us the following correlation matrix visualization:

And also to highlight numbers:

1. # Highlight largest number in column:
   
2. 
   
3. df.style.highlight_max(axis=0)
   

复制代码

The above coding gives us the following correlation matrix visualization:

It is also possible to highlight large numbers and color negative number red with the following code:

1. #Color negative numbers red
   
2. 
   
3. def color_negative_red(val):
   
4. 
   
5.     """
   
6. 
   
7.     Takes a scalar and returns a string with
   
8. 
   
9.     the css property `'color: red'` for negative
   
10. 
    
11.     strings, black otherwise.
    
12. 
    
13.     """
    
14. 
    
15.     color = 'red' if val < 0 else 'black'
    
16. 
    
17.     return 'color: %s' % color
    
18. 
    
19. 
    
20. 
    
21. # Hightlight large numbers
    
22. 
    
23. def highlight_max(s):
    
24. 
    
25.     '''
    
26. 
    
27.     highlight the maximum in a Series yellow.
    
28. 
    
29.     '''
    
30. 
    
31.     is_max = s == s.max()
    
32. 
    
33.     return ['background-color: yellow' if v else '' for v in is_max]
    
34. 
    
35. # Create correlation matrix:
    
36. 
    
37. rs = np.random.RandomState(0)
    
38. 
    
39. df = pd.DataFrame(rs.rand(10, 10))
    
40. 
    
41. corr = df.corr()
    
42. 
    
43. corr.style.\
    
44. 
    
45.     applymap(color_negative_red).\
    
46. 
    
47.     apply(highlight_max)
    

复制代码

The above coding gives us the following correlation matrix:

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏1 回帖

关键词：correlation Matrices relation matrice matric