Visualizing common drawings （ 5、 ... and ） Scatter plot

One . Introduction to scatter diagram

Scatter diagram is also called X-Y chart , It shows all the data in the form of points in the rectangular coordinate system , To show the degree of interaction between variables , The position of the point is determined by the value of the variable .

By observing the distribution of data points on the scatter plot , We can infer the correlation between variables . If there is no correlation between variables , Then, in the scatter diagram, it will be shown as randomly distributed discrete points , If there's a correlation , Then most of the data points will be relatively dense and present a certain trend . The correlation of data is mainly divided into ：

• positive correlation （ The values of two variables increase at the same time ）.
• negative correlation （ The value of one variable increases and the value of the other decreases ）.
• Unrelated .
• Linear correlation .
• Exponential correlation .

Scatter charts are often used in conjunction with regression lines , Summarize and analyze the existing data for prediction analysis .

For those variables, there is a close relationship , But these relationships are not as accurate as mathematical and physical formulas , Scatter chart is a good graphic tool . But in the analysis process, we need to pay attention to , The correlation between these two variables is not equivalent to a definite causal relationship , Other influencing factors may also need to be considered .

Two . Composition of scatter diagram

A standard scatter diagram includes at least the following parts ：

• The vertical axis ： Represents the value of one of the variables
• The horizontal axis ： Represents the value of one of the variables
• spot ：（X,Y）
• The regression line ： The line that runs through all points most accurately

3、 ... and . Application scenarios

Fit data ： Data from two consecutive data fields .

The main function ： Observe the distribution of data .

Number of applicable data ： unlimited .

remarks ： In order to better observe the data distribution , You need to set the transparency or color of data points .

Suitable for the scene ：

• Display and compare values , Not only can it show trends , It can also display the shape of the data cluster , And the relationship of data points in the data cloud .

Not suitable for the scene ：

• Display the proportion of each classification data .

Four . Realization

stay matplotlib Use in scatter Function to realize scatter diagram , The functions are described as follows ：

scatter(x, y, s=None, c=None, marker=None, cmap=None, norm=None,vmin=None, vmax=None, alpha=None, linewidths=None, *,edgecolors=None, plotnonfinite=False, data=None, **kwargs)

Parameters 1：x,y： Specify the coordinates of the data scatter .

Parameters 2：s： Numerical type , Specifies the size of the scatter .

Parameters 3：c： Array or class array type , Specifies the color of the scatter .

Parameters 4：marker： Qualified string , Specifies the marker type of the scatter （ The default is ：‘o’）.

Parameters 5：cmap： Specify the selected colormap.

Parameters 6：norm： Unknown .

Parameters 7、8：min、vmax and norm Used together to normalize data .

Parameters 9：alpha： floating-point , Specifies the transparency of the scatter .

Parameters 10：linewidths： Integer type , Specifies the lineweight of the scatter edge ; If marker by None, Then use verts To construct a scatter marker

Parameters 11：verts： Unknown .

Parameters 12：edgecolors： Array or class array type , Specifies the scatter edge color , Will cycle .

Parameters 13：plotnonfinite： Boolean type , combination set_bad Use , Specifies whether to draw points in an unrestricted way .

Parameters 14：**kwargs： The accepted keyword parameters are passed to Collection example .

Return value ： The associated PathCollection example .

Use to SOCR-HeightWeight.csv Data sets, for example , The data set records a total of 25000 The height and weight of an object , With height as the horizontal axis , Take weight as the vertical axis , Look at the relationship between two variables , The complete code is as follows ：

import matplotlib.pyplot as plt

from sklearn.linear_model import LinearRegression

import numpy as np

import pandas as pd

plt.rcParams['font.sans-serif'] = ['SimHei']  #  Settings support Chinese 
plt.rcParams['axes.unicode_minus'] = False  #  Set up - Number 
plt.style.use('seaborn-dark-palette')

df = pd.read_csv("SOCR-HeightWeight.csv", index_col=0)
height = df["Height(Inches)"].values.reshape(-1, 1)
weight = df["Weight(Pounds)"].values.reshape(-1, 1)

model = LinearRegression()
model.fit(height, weight)
coef = model.coef_[0]
intercept = model.intercept_[0]

height_avg = np.average(height)
weight_avg = np.average(weight)

quadrant1 = df[(df["Height(Inches)"] >= height_avg) & (df["Weight(Pounds)"] >= weight_avg)]
quadrant1_height = quadrant1["Height(Inches)"][:3000]
quadrant1_weight = quadrant1["Weight(Pounds)"][:3000]
plt.scatter(quadrant1_height, quadrant1_weight, alpha=0.3, label=" Scatter plot first quadrant ")

quadrant2 = df[(df["Height(Inches)"] <= height_avg) & (df["Weight(Pounds)"] >= weight_avg)]
quadrant2_height = quadrant2["Height(Inches)"][:3000]
quadrant2_weight = quadrant2["Weight(Pounds)"][:3000]
plt.scatter(quadrant2_height, quadrant2_weight, alpha=0.3, label=" Scatter plot second quadrant ")

quadrant3 = df[(df["Height(Inches)"] <= height_avg) & (df["Weight(Pounds)"] <= weight_avg)]
quadrant3_height = quadrant3["Height(Inches)"][:3000]
quadrant3_weight = quadrant3["Weight(Pounds)"][:3000]
plt.scatter(quadrant3_height, quadrant3_weight, alpha=0.3, label=" The third quadrant of the scatter chart ")

quadrant4 = df[(df["Height(Inches)"] >= height_avg) & (df["Weight(Pounds)"] <= weight_avg)]
quadrant4_height = quadrant4["Height(Inches)"][:3000]
quadrant4_weight = quadrant4["Weight(Pounds)"][:3000]
plt.scatter(quadrant4_height, quadrant4_weight, alpha=0.3, label=" The fourth quadrant of the scatter chart ")

#  Draw average 
plt.hlines(weight_avg, min(height), max(height), ls="--", color='r', lw=2, label=' Average weight ')
plt.vlines(height_avg, min(weight), max(weight), ls='--', color='k', lw=2, label=' Average height ')
x = np.arange(min(height), max(height), 0.05)
y = coef * x + intercept
plt.plot(x, y, lw=2, color="darkgray", label=" Regression line of height and weight ")
plt.title(" Scatter diagram of height and weight ", fontsize=25, fontweight="bold")
plt.xlabel(" height (Inches)", fontsize=20)
plt.ylabel(" weight (Pounds)", fontsize=20)
plt.legend(fontsize=15)
plt.show()

The effect is as follows ：

5、 ... and . Reference resources

1. Introduction to stacking diagram
2. Broken line diagram
3. Area map
4. Histogram