Core Steps for Python Visualization

In regular research, the three steps to achieve visualization using Python are:

Identify the problem and choose the graph
Transform the data and apply functions
Set parameters to make it clear

1. First, which libraries do we use for plotting?matplotlibis the most basic plotting library in Python, and it is generally the starting point for Python data visualization before moving on to vertical and horizontal expansions.Seabornis a high-level visualization library based on matplotlib, mainly targeting variable feature selection in data mining and machine learning. Seaborn can create visualizations of multi-dimensional data with concise code.Other libraries includeBokeh (a library for interactive visualizations on the browser, enabling interaction between analysts and data);Mapbox (a more powerful visualization tool library for handling geographic data) etc.This article mainly uses matplotlib for case analysisStep 1: Identify the problem and choose the graphThe business may be complex, but through decomposition, we need to find what specific problem we want to express through the graph. Training in analytical thinking can learn from the methods in The McKinsey Method and The Pyramid Principle.This is a summary of chart type selection found online. Core Steps for Python Visualization In Python, we can summarize the following four basic visual elements to present graphs:

Points: scatter plot for two-dimensional data, suitable for simple two-dimensional relationships;
Lines: line plot for two-dimensional data, suitable for time series;
Bars: bar plot for two-dimensional data, suitable for categorical statistics;
Colors: heatmap suitable for displaying a third dimension;

There are relationships between data such as distribution, composition, comparison, connection, and trend changes. Choose the corresponding graph to display according to different relationships.Step 2: Transform the data and apply functionsMuch of the programming work in data analysis and modeling is based on data preparation: loading, cleaning, transforming, and reshaping. Our visualization steps also need to organize the data, transforming it into the format we need before applying visualization methods to complete the plotting.Here are some common data transformation methods:

Merge: merge, concat, combine_first (similar to a full outer join in databases)
Reshape: reshape; pivot (similar to Excel pivot tables)
Remove duplicates: drop_duplicates
Map: map
Fill and replace: fillna, replace
Rename axis indexes: rename

Convert categorical variables to ‘dummy variable matrix’ using the get_dummies function and limit values on a column in df, etc.Functions are then chosen based on the graph selected in the first step, to find the corresponding functions in Python.Step 3: Set parameters to make it clearAfter the original graph is plotted, we can modify colors (color), line styles (linestyle), markers (marker), or other chart decoration items such as titles (Title), axis labels (xlabel, ylabel), axis ticks (set_xticks), and legends (legend) to make the graph more intuitive.The third step is based on the second step, to make the graph clearer and more understandable, and is a work of embellishment. Specific parameters can be found in the plotting function.2. Basics of Visualization PlottingMatplotlib Plotting Basics

# Import packages
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

Figure and SubplotAll matplotlib graphics are located in Figure (canvas), Subplot creates image space. You cannot plot through figure; you must create one or more subplots using add_subplot.figsize can specify image size.

# Create canvas
fig = plt.figure()
<figure>
# Create subplot, 221 indicates this is the first image in a 2x2 grid.
ax1 = fig.add_subplot(221)
# But now it is more common to create canvas and images as follows, 2,2 indicates this is a 2*2 canvas that can hold 4 images
fig , axes = plt.subplots(2,2,sharex=True,sharey=True)
# The sharex and sharey parameters of plt.subplot can specify that all subplots use the same x and y axis ticks.</figure>

You can adjust spacing using the subplots_adjust method of Figure.

subplots_adjust(left=None,bottom=None,right=None,top=None,wspace=None,hspace=None)

Color, Marker, and Line StyleThe plot function of matplotlib accepts a set of X and Y coordinates and can also accept a string abbreviation representing color and line style: ‘g–‘ indicates the color is green and the line style is ‘–‘ dashed. You can also specify parameters explicitly.Line plots can also add some markers (marker) to highlight the position of data points. Markers can also be included in the format string, but the marker type and line style must be placed after the color.

plt.plot(np.random.randn(30),color='g',linestyle='--',marker='o')
[]

Ticks, Labels, and LegendsThe plt’s xlim, xticks, and xtickslabels methods control the range and tick positions and tick labels of the chart.When called without parameters, it returns the current parameter values; when called with parameters, it sets the parameter values.

plt.plot(np.random.randn(30),color='g',linestyle='--',marker='o')
plt.xlim() # Call without parameters to display current parameters;
# You can replace xlim with the other two methods and try
(-1.4500000000000002, 30.45)

plt.plot(np.random.randn(30),color='g',linestyle='--',marker='o')
plt.xlim([0,15]) # Change x-axis ticks to 0-15
(0, 15)

Set Title, Axis Labels, Ticks, and Tick Labels

fig = plt.figure();ax = fig.add_subplot(1,1,1)
ax.plot(np.random.randn(1000).cumsum())
ticks = ax.set_xticks([0,250,500,750,1000]) # Set tick values
labels = ax.set_xticklabels(['one','two','three','four','five']) # Set tick labels
ax.set_title('My first Plot') # Set title
ax.set_xlabel('Stage') # Set axis label
Text(0.5,0,'Stage')

Add LegendThe legend is another important tool for identifying chart elements.You can pass the label parameter when adding a subplot.

fig = plt.figure(figsize=(12,5));ax = fig.add_subplot(111)
ax.plot(np.random.randn(1000).cumsum(),'k',label='one') # Pass in label parameter to define label name
ax.plot(np.random.randn(1000).cumsum(),'k--',label='two')
ax.plot(np.random.randn(1000).cumsum(),'k.',label='three')
# After the graph is created, just call the legend parameter to bring out the label.
ax.legend(loc='best') # If the requirement is not strict, it is recommended to use loc='best' parameter to let it choose the best position itself

AnnotationsIn addition to standard chart objects, we can also customize adding text annotations or arrows.Annotations can be added using functions like text, arrow, and annotate. The text function can draw text at specified x, y coordinates, and can also be customized.

plt.plot(np.random.randn(1000).cumsum())
plt.text(600,10,'test ',family='monospace',fontsize=10)
# Chinese annotations cannot display normally in the default environment, you need to modify the configuration file to support Chinese fonts. Please search for specific steps yourself.

Save the Chart to a FileYou can use plt.savefig to save the current chart to a file. For example, to save the chart as a png file, you can execute:The file type is determined by the extension. Other parameters include:

fname: A string containing the file path, the extension specifies the file type
dpi: Resolution, default 100 facecolor, edgecolor the background color of the image, default ‘w’ white
format: Display settings for file format (‘png’, ‘pdf’, ‘svg’, ‘ps’, ‘jpg’, etc.)
bbox_inches: The part of the chart that needs to be retained. If set to ‘tight’, it will try to cut out the blank parts around the image

plt.savefig('./plot.jpg') # Save the image as a jpg format image named plot
<figure></figure>

3. Plotting Functions in Pandas

Matplotlib Plottingmatplotlib is the most basic plotting function and is also relatively low-level.To assemble a chart, you need to call each basic component separately. Pandas has many advanced plotting methods based on matplotlib, and charts that originally required multiple lines of code can be done in just a few lines using pandas.What we use calls the plotting package in pandas.

import matplotlib.pyplot as plt

Line PlotBoth Series and DataFrame have a plot method for generating various charts.By default, they generate line plots.

s = pd.Series(np.random.randn(10).cumsum(),index=np.arange(0,100,10))
s.plot() # The index of the Series object will be passed to matplotlib as the x-axis for plotting.

df = pd.DataFrame(np.random.randn(10,4).cumsum(0),columns=['A','B','C','D'])
df.plot() # plot will automatically change colors for different variables and add legends

Parameters of Series.plot Method

label: For chart labels
style: Style string, ‘g–‘
alpha: Image fill opacity (0-1)
kind: Chart type (bar, line, hist, kde, etc.)
xticks: Set x-axis tick values
yticks: Set y-axis tick values
xlim, ylim: Set axis limits, [0,10]
grid: Show axis grid lines, default off
rot: Rotate tick labels
use_index: Use the object’s index as tick labels
logy: Use logarithmic scale on Y-axis

Parameters of DataFrame.plot MethodIn addition to the parameters in Series, DataFrame has some unique options.

subplots: Plot each DataFrame column in separate subplots
sharex, sharey: Share x and y axes
figsize: Control image size
title: Image title
legend: Add legend, default display
sort_columns: Draw columns in alphabetical order, default uses current order

Bar ChartBy adding kind=’bar’ or kind=’barh’ to the code for generating line charts,you can generate bar charts or horizontal bar charts.

fig,axes = plt.subplots(2,1)
data = pd.Series(np.random.rand(10),index=list('abcdefghij'))
data.plot(kind='bar',ax=axes[0],rot=0,alpha=0.3)
data.plot(kind='barh',ax=axes[1],grid=True)

Bar charts have a very practical method:Use value_counts to visually display the frequency of occurrences of each value in a Series or DF.For example, df.value_counts().plot(kind=’bar’)The basic syntax for Python visualization ends here, and other graph plotting methods are largely similar.The key is to follow the thinking of the three steps to think, choose, and apply. More practice can make you more proficient.

This article is sourced from: https://blog.csdn.net/m0_72557783/article/details/125623698. It is only for passing and sharing more information and does not represent the platform's endorsement of its views or its accuracy. Copyright belongs to the original author. If there is any infringement, please contact us for deletion.

Editor / Zhang Zhihong

Review / Fan Ruiqiang

Verification / Zhang Zhihong

Reprinted from: Mathematics China