Visualization of Python data analysis-summarized with an article: matplotlib, Pandas, seaborn

Python's data visualization tools mainly rely on matplotlib, pandas and seaborn.

1. Use matplotlib for data visualization

1.1 Basic concepts

• Canvas/Painting Board
• Coordinate system/drawing interval
  

1.2 Core Steps: Three Steps to Drawing a Picture

Take line chart as an example

1. Define coordinate points (prepare data)
2. plot(x, y) Draw a graph (the default is a line graph)
3. plt.show() Display graph
   Note: No matter how many, x and y in the plot correspond one-to-one and appear in pairs .

1.3 Detailed introduction:

Remember:
if you don’t specify the 画板figure()sum 子图subplot, a drawing board and a subgraph will be created by defaultfigure(1)subplot(1,1,1)

1. Build the canvas

Create canvas + set canvas size : plt.figure()
you can also pass in the canvas size parameter figsize = (8, 6)to adjust the canvas size!

2. Establish a coordinate system (determine the drawing area)

There are several different methods:

2.1 Canvas segmentation + return to all coordinate systems

plt.subplots()

The method subsequent passes axes[x,y]is plotted to indicate a coordinate system which

2.2 Canvas segmentation + designated coordinate system position (to return)

ax= fig.add_subplot()
plt.subplot2grid()
plt.subplot()

The first method belongs to object programming, the latter three belong to functional programming

The following three code examples:

• plt.subplot2grid()

plt.subplot2grid((2,2),(0,0)) 
plt.subplot2grid((2,2),(0,1)) 
plt.subplot2grid((2,2),(1,1)) 

Control the position of the coordinate system by coordinates

• subplot()

plt.subplot(2,2,1)
plt.subplot(2,2,2)
plt.subplot(2,2,4)

By digital control of position coordinates

combat:

for i in range(len(crops)):  # crops 为 10 张堆叠的图片 , 大小：(10, 224, 224, 3)
    plt.subplot(2,5,i+1)   
    plt.imshow(crops[i, :, :, :])

result:

• subplots()

plt.subplots(2,2)

Return all (2x2) coordinate systems

3. Set the coordinate axis

Set the title of the axis

plt.xlabel("str")
plt.ylabel("str")
plt.title('str')

Parameters labelpadmay also set the title to axis distance;
other input parameters can stringbe set

Set the scale of the axis

Customize which scales to display the scale value

plt.xticks(ticks,labels)
plt.yticks(ticks,labels)

Tip: You can hide the values ​​of the x/y axis
by passing in an empty list to ensure data security .

plt.xticks([])
plt.yticks([])

Set the range of the coordinate axis

plt.xlim()
plt.ylim()

You can directly pass in the two numbers of the starting point and the ending point as parameters.

Or use a simpler method:,
axis[xmin, xmax, ymin, ymax]for example,

plt.axis([0, 6, 0, 20])

Note that although the input here is in the form of a list, it will actually be converted to a numpy array form internally to make it easier for us to process the data.

other settings

-Turn off the axis display:plt.axis('off')

-Open grid line: plt.grid(b = 'True')
also can be passed in axis参数, specify to open only the specified axis

- the legend
on plt.plot()the incoming label parameters , such as the label = ‘str’
then plt.legend()displayed
...

5. Draw a chart

-Line chart: plt.plot(x,y)
-Histogram: plt.bar(x,y)
-Scatter chart: -Heat plt.scatter(x,y)
map: plt.imshow(x,cmap)
…

• Optional parameters used in drawing 2d diagrams:
  

6. Icon display

plt.show()

1.4 Frequently Asked Questions

• Is to solve the problem of not displaying the image

%matplotlib inline

• Solve Chinese garbled

plt.rcParams['font.sans-serif']='SimHei'

1.5 Minimal code implementation

import numpy as np
import matplotlib.pyplot as plt

A coordinate system

As mentioned earlier, some operations can be omitted

x = np.linspace(1, 20, 100)

y1 = np.sin(x)
y2 = np.cos(x)

plt.plot(x, y1)
plt.plot(x, y2)

Multiple coordinate systems

x = np.linspace(1, 20, 100)

y1 = np.sin(x)
y2 = np.cos(x)

plt.subplot(2,2,1)
plt.plot(x, y1)
plt.subplot(2,2,4)
plt.plot(x, y2)

2. Use Pandas for data visualization

PandasThe drawing is a matplotlibpackage made on the
basic syntax is:
df.plot(x='列名1', y='列名2', kind='图形类型', label=‘图例名称’)

line graph

from numpy.random import randn
np.random.seed(1)

df = pd.DataFrame(np.random.randn(20,3),index=np.linspace(0,19,20), columns=list('ABC'))

df.plot()

Bar graph

from numpy.random import randn
np.random.seed(1)

df = pd.DataFrame(np.random.randn(5,3)+10,index=np.linspace(0,4,5), columns=list('ABC'))

df.plot.bar()

Histogram

from numpy.random import randn
np.random.seed(1)

df = pd.DataFrame({
    
    'A':np.random.randn(100),'B':np.random.randn(100)+1,'C':np.random.randn(100)+2})

df.hist(bins=20)

Box plot

from numpy.random import randn
np.random.seed(1)

df = pd.DataFrame(np.random.rand(10, 5), columns=['A', 'B', 'C', 'D', 'E'])
df.plot.box()

Scatter plot

from numpy.random import randn
np.random.seed(1)

df = pd.DataFrame(np.random.rand(50, 4), columns=['a', 'b', 'c', 'd'])
df.plot.scatter(x='a', y='b')

Pie chart

from numpy.random import randn
np.random.seed(1)

df = pd.DataFrame(3 * np.random.rand(4), index=['a', 'b', 'c', 'd'], columns=['x'])
df.plot.pie(subplots=True)

3. Use seaborn for data visualization

Use Seaborn for data visualization

to sum up

• One-dimensional graph:
  (not directly meaningful) one-dimensional data
  box plot
• Two-dimensional graphs
  Scatter graphs , line graphs, histograms, bar graphs
• Three-dimensional diagram:
  bubble diagram

To be continued~

Attachment: the choice of visual graphics

Visual Basics: How to Choose Visual Graphics