Visual introduction to Matplotlib and plotnine

  Matplotlib yes Python in similar MATLAB Drawing tools for , The author often uses Matlab, So think of it as an introduction to data visualization , But its style is often more regular , The grammar is also more complex to adjust .Plotnine It makes up for Python Lack of data , similar R Language , Can be said to be ggplot2 stay Python Porting version on , The author had a preliminary contact with the American competition because of its exquisite drawing style , It is easier to adjust the grammar , But the Chinese language is not supported , There is also less relevant information . Now take advantage of the summer semester , Introduce their basic usage .

The architecture

   about Matplotlib for , What it does is a serial addition , have Layer upon layer Characteristics . At the bottom Canvas( Drawing board ) , Build on Figure( canvas ) , And then build on the canvas Axes( Drawing area ) , And the axis (axis) 、 legend (legend) And other auxiliary display layers and image layers are built on Axes above , Drawing object oriented .

   This is very similar to the process of drawing with a drawing board , Let's find a drawing board first , Then lay a canvas , Select the painting area on the canvas , We painted blue sky and white clouds in this area , And can continue to write beside the blue sky and white clouds .

   about Plotnine for , What it does is a parallel addition , have Step by step Characteristics . This graphical syntax divides the drawing process into data (data)、 mapping (mapping)、 Geometry object (geom)、 Statistical transformation (stats)、 scale (scale)、 Coordinate system (coord)、 Split up (facet) Several parts , The parts are independent of each other . differ Matplotlib, When there is a strong correlation between before and after , Changes often become cumbersome , and Plotnine Just modify the settings of the corresponding part .

   It's very much like the process of decorating a statue , Add a garland , Wear a hat , But it's just as easy to remove the wreath and hat from the statue , Especially the style theme (theme) Selection of , Make style adjustment more convenient .

Plot function

   about Matplotlib Come on , The drawing function is called matplotlib.pyplot, It is generally recorded as plt; about Plotnine for , yes ggplot. Make good use of these two basic functions , Then find the type of chart we need to draw , Can complete the basic drawing task . Take an example , For example, we now have such a set of data , To draw a histogram ：

median_age_dict = {
    
    'coutry': ['New Zealand', 'Spain', 'Ireland', 'Israel', 'Denmark', 'Norway', 'Netherlands', 'Australia', 'Italy', 'Sweden'],
    'age': [39.0, 37.0, 35.0, 34.0, 34.0, 34.0, 34.0, 34.0, 34.0, 34.0]
    }

plt

import matplotlib.pyplot as plt
import pandas as pd

median_age_dict = {
    
    'country': ['New Zealand', 'Spain', 'Ireland', 'Israel', 'Denmark', 'Norway', 'Netherlands', 'Australia', 'Italy',
           'Sweden'],
    'age': [39.0, 37.0, 35.0, 34.0, 34.0, 34.0, 34.0, 34.0, 34.0, 34.0]
}
median_age = pd.DataFrame(median_age_dict)

#  Create a form 
plt.figure('matplotlib')
#  mapping 
plt.bar(median_age['country'],median_age['age'])
#  Exhibition 
plt.show()

ggplot

from plotnine import *
import pandas as pd

median_age_dict = {
    
    'country': ['New Zealand', 'Spain', 'Ireland', 'Israel', 'Denmark', 'Norway', 'Netherlands', 'Australia', 'Italy',
           'Sweden'],
    'age': [39.0, 37.0, 35.0, 34.0, 34.0, 34.0, 34.0, 34.0, 34.0, 34.0]
}
median_age = pd.DataFrame(median_age_dict)

#  mapping , a fly in the ointment , No Chinese support 
sep = (ggplot(median_age,aes('country','age',fill='country'))+
              #  Two dimensional bar chart 
              geom_col()
              )
#  Exhibition 
print(sep)

   I can see you ,ggplot The drawing style of is more bright , When directly displayed, the elements of drawing display are more abundant , and plt Is more regular , It belongs to the state that what is added can be displayed , The legend and axis names are not actively displayed .

Common sense of style diversity

   Master the histogram 、 Drawing of curve and scatter diagram , Know how to adjust the legend and coordinate axis , Able to adjust line type and point type , And can complete the specified form of image saving , It can basically meet the needs of drawing in daily learning . You can specify separately when you are free in the future .

   about Matplotlib Learning from , matplotlib Official website The above information is sufficient , There are also many detailed information among the people , Such as Matplotlib course , Its grammar and Matlab The approximate , Therefore, it can also be used with Matlab Analogical transfer of the use of functions in .

   about Plotnine Learning from ,《python The beauty of data visualization 》 A good example map is provided , It is shown below . You can also refer to R Linguistic ggplot2 To learn grammar , There is also a lot of folk information about this , Such as ggplot2 Efficient and practical guide .

Plotnine Learning manual

Matplotlib Simulated dynamic forest fire

   Finally, I will show you two more complex examples that I have done , The first is an example of Simulating Forest Dynamic fire , use Matplotlib Draw a dynamic display of the evolution of the square .

#Here we have a 2-D battle field
#One side is green, the other is blue
import matplotlib.animation as animation  #  Save animation 
import matplotlib.pyplot as plt
import numpy as np
from matplotlib.ticker import MultipleLocator
import random

#define class cell
class Cell:

    def __init__(self,HP,GEN,site):
        self.HP = HP
        #self.POS = POS
        self.GEN = GEN
        self.site = site

#The LOS caused by the enemies
    def Attack(self):
        LOS =  random.uniform(2, 4)
        return LOS

#The CURE provided by the teammates
    def Heal(self):
        CURE = random.uniform(1,3)
        return CURE

#Judge if the cell is dead
    def Sign(self):
        if self.HP <= 0:
            return True
        else:
            return False

#create the cell list
def create_cells(Num,HP):
    # the list of sites
    sites = []
    for j in range(int(np.sqrt(Num))):
        for k in range(int(np.sqrt(Num))):
            sites.append([j, k])
    random.shuffle(sites)

    # The amounts of green cells
    Green = random.randint(int(np.fix(np.sqrt(Num))), Num)
    cells = []
    for i in range(0, Green):
        cells.append(Cell(HP[0], 'green', sites[i]))
    for i in range(Green, Num):
        cells.append(Cell(HP[1], 'blue', sites[i]))
    return cells

#the battle between two cells
def battle(Cell_1,Cell_2):
    if Cell_1.GEN == Cell_2.GEN:
        Cell_1.HP += Cell_2.Heal()
        Cell_2.HP += Cell_1.Heal()
    else:
        Cell_1.HP -= Cell_2.Attack()
        Cell_2.HP -= Cell_1.Attack()
    return [Cell_1.HP,Cell_2.HP]

#the war between the cells
def war(cells):
    # for i in range(len(cells)):
    # HP = battle(cells[i-1], cells[i])
    # cells[i-1].HP = HP[0]
    # cells[i].HP = HP[1]
    Num = len(cells)
    Grid = int(np.sqrt(Num))
    for cell_1 in cells:
        for cell_2 in cells:
            if cell_1.site[1] == cell_2.site[1]:
                if cell_1.site[0] == 0:
                    if cell_2.site[0] == 1 or cell_2.site[0] == Grid:
                        HP = battle(cell_1, cell_2)
                        cell_1.HP = HP[0]
                        cell_2.HP = HP[1]
                elif cell_1.site[0] == Grid:
                    if cell_2.site[0] == 0 or cell_2.site[0] == Grid - 1:
                        HP = battle(cell_1, cell_2)
                        cell_1.HP = HP[0]
                        cell_2.HP = HP[1]
                else:
                    if cell_2.site[0] == cell_1.site[1] - 1 or cell_2.site[0] == cell_1.site[1] + 1:
                        HP = battle(cell_1, cell_2)
                        cell_1.HP = HP[0]
                        cell_2.HP = HP[1]
            elif cell_1.site[0] == cell_2.site[0]:
                if cell_1.site[1] == 0:
                    if cell_2.site[1] == 1 or cell_2.site[1] == Grid:
                        HP = battle(cell_1, cell_2)
                        cell_1.HP = HP[0]
                        cell_2.HP = HP[1]
                elif cell_1.site[1] == Grid:
                    if cell_2.site[1] == 0 or cell_2.site[1] == Grid - 1:
                        HP = battle(cell_1, cell_2)
                        cell_1.HP = HP[0]
                        cell_2.HP = HP[1]
                else:
                    if cell_2.site[1] == cell_1.site[1] - 1 or cell_2.site[1] == cell_1.site[1] + 1:
                        HP = battle(cell_1, cell_2)
                        cell_1.HP = HP[0]
                        cell_2.HP = HP[1]

    return cells

# visual the war
def visual(cells,ims):
    ax = plt.subplot(111)
    plt.ion()
    reddead = 0
    greenlive = 0
    bluelive = 0

    # get the color
    for cell in cells:
        x = np.linspace(cell.site[0], cell.site[0] + 1)
        if cell.Sign():
            ax.fill_between(x, cell.site[1], cell.site[1] + 1, facecolor='red')
            reddead += 1
        else:
            if cell.GEN == 'green':
                ax.fill_between(x, cell.site[1], cell.site[1] + 1, facecolor='green')
                greenlive += 1
            elif cell.GEN == 'blue':
                ax.fill_between(x, cell.site[1], cell.site[1] + 1, facecolor='blue')
                bluelive += 1

    # set the scale
    plt.xlim(0, np.sqrt(len(cells)))
    plt.ylim(0, np.sqrt(len(cells)))
    ax.xaxis.set_major_locator(MultipleLocator(1))
    ax.yaxis.set_major_locator(MultipleLocator(1))
    ax.xaxis.grid(True, which='major')  # major,color='black'
    ax.yaxis.grid(True, which='major')  # major,color='black'
    plt.pause(0.8)
    ims.append(ax.findobj())

    # print([reddead, greenlive, bluelive])
    return [reddead, greenlive, bluelive]

def simulata(cells):
     fig = plt.figure()
     ims = []
     initial = visual(cells,ims)
     state = [initial]
     i = 0
     while state[i][0] < initial[1] and state[i][0] < initial[2]:
         cells = war(cells)
         state.append(visual(cells, ims))
         i +=1
     ani = animation.ArtistAnimation(fig, ims, interval=500, repeat_delay=1000)
     ani.save("test.gif", writer='pillow')

     return state


# Here we have a try, assume that we have Num cells
Num = 256
cells = create_cells(Num,[10,10])
state = simulata(cells)
print('Below is the result of simulation:\n',state)
greenlive_ratio = state[-1][1]/state[0][1]
bluelive_ratio = state[-1][2]/state[0][2]
print('\ngreenlive ratio is ',state[-1][1],'/',state[0][1], greenlive_ratio,
      '\nbluelive ratio is',state[-1][2],'/',state[0][2],bluelive_ratio)

The effect display is a moving picture , And by the matplotlib.animation Save the dynamic diagram .

Plotnine Point line diagram

The other is to use Plotnine A plot of a set of data , At this time, you can see that the whole style is compact and elegant .

import numpy as np
from plotnine import *
import pandas as pd

def pit():
	#  data 
    df = {
    'x': [], 'y': [], 'label': []}

    y1 = [25978.40, 25978.40, 25978.40, 25978.40, 25978.40, 25978.40,
          25978.40, 25978.40, 25978.40, 25978.40, 25978.40]

    y2 = [12511.43, 12511.43, 12511.43, 12511.43, 12511.43, 11440.20,

          10368.96, 14418.96, 18468.96, 24969.96, 31470.96]

    y3 = [47111.58, 47111.58, 47111.58, 46111.58, 45111.58, 42602.52,

          40093.45, 35093.45, 30093.45, 27264.94, 24436.42]

    y4 = [7055.95, 7055.95, 7055.95, 7555.95, 8055.95, 9550.86,
          11045.76, 8074.72, 5103.67, 4645.83, 4187.98]

    y5 = [86468.18, 86468.18, 86468.18, 86968.18, 87468.18, 89553.58,

          91638.97, 95560.02, 99481.06, 96266.42, 93051.78]

    Num = 11
    for i in range(Num):
        df['x'].append(i + 1)
        df['y'].append(y1[i])
        df['label'].append('CO')

    for i in range(Num):
        df['x'].append(i + 1)
        df['y'].append(y2[i])
        df['label'].append('WY')

    for i in range(Num):
        df['x'].append(i + 1)
        df['y'].append(y3[i])
        df['label'].append('AZ')

    for i in range(Num):
        df['x'].append(i + 1)
        df['y'].append(y4[i])
        df['label'].append('NM')

    for i in range(Num):
        df['x'].append(i + 1)
        df['y'].append(y5[i])
        df['label'].append('CA')

    cases = ['(0,1)', '(0.2,0.8)', '(0.4,0.6)', '(0.6,0.4)', '(0.8,0.2)', '(1,0)']

    df = pd.DataFrame(df)

    COLOR = ('#F91400', '#FF328E')
    SHAPE = ('o', 's')

    #  mapping 
    first_plot = (ggplot(df, aes('x', 'y', shape='label', fill='label', color='label')) +
    			 #  Combination of dot line diagram  
                  geom_point(size=3) +
                  geom_line(size=1) +
                  labs(x=' ', y="", title=' ') +
                  #  Axis adjustment 
                  scale_x_continuous(limits=(0, 11), breaks=np.linspace(1, 11, 6), labels=cases) +
                  #  Style adjustment , This is a very convenient and flexible step 
                  theme_bw(
                  )
                  )

    first_plot.save('alplan.png', width=10, height=4.5)
    print(first_plot)

pit()