Y轴Matplotlib上的双标签

Question

I've made a barh graph with a scatter-plot on top. The data is about 100 books and the publishing date along with the year the author was born and died. The barh shows the time the author was alive and the scatter-plot shows the year there books where published.

The problem I am facing is being able to plot multiple books on one bar. As I have duplicate bars now with different books. I am creating the y-axis based on position in the array and I'm adding the label later.

My relevant code:

# dataframe columns to arrays. (dataset is my pandas dataframe)
begin = np.array(dataset.BORN)
end = np.array(dataset.DIED)
book = np.array(dataset['YEAR (BOOK)'])

# Data to a barh graph (sideways bar)
plt.barh(range(len(begin)), end-begin, left=begin, zorder=2, 
color='#007acc', alpha=0.8, linewidth=5)

# Plots the books in a scatterplot. Changes marker color and shape.
plt.scatter(book, range(len(begin)), color='purple', s=30, marker='D', zorder=3)

# Sets the titles of the y-axis.
plt.yticks(range(len(begin)), dataset.AUTHOR)

# Sets start and end of the x-axis.
plt.xlim([1835, 2019])

# Shows the plt
plt.show()

Picture that shows part of my current graph:

Answer 1

I'd aggregate your dataset down so that you get a single author per row using groupby and use this to draw the bars, then join this back to get a value to use to draw the books, eg:

import pandas as pd
import matplotlib.pyplot as plt

df = pd.DataFrame([
    ['foo', 1950, 1990, 1980],
    ['foo', 1950, 1990, 1985],
    ['bar', 1930, 2000, 1970],
], columns=['author', 'born', 'died', 'published'])

pulls in packages and creates a dummy dataset, next we reduce this down to a single row per author, getting when they were born & died:

agg = df.groupby('author')['born', 'died'].agg(min).reset_index()
agg['auth_num'] = range(len(agg))

the reset_index makes the author back into a normal column, and we create an arbitrary auth_num column, you might want to put a sort_values in there if you want to sort authors by something other than their name (which I'd recommend as alphabetical generally isn't the most useful )

next we can join this back on to the original dataset to get an author number for each book:

df2 = pd.merge(df, agg[['author', 'auth_num']], on='author')

and finally plot it all:

plt.barh(agg.auth_num, agg.died - agg.born, left=agg.born, zorder=-1, alpha=0.5)
plt.yticks(agg.auth_num, agg.author)

plt.scatter(df2.published, df2.auth_num)

giving something like:

note: if you set use_sticky_edges to False before calling barh , it'll allow the x-axis to auto-scale and hence the left-most author won't "stick" to the left-hand margin

Answer 2

Sure, there are several options you could use. You could either create another array for 1st, 2nd, 3rd books. Or you could create a dictionary or list of arrays to plot the books per author.

I have re-produced some examples using dummy data below.

import matplotlib.pyplot as plt
import numpy as np

fig,axs = plt.subplots(1,1,figsize=(10,10))

# dataframe columns to arrays. (dataset is my pandas dataframe)
begin = np.arange(1900,1950)
end = np.arange(1975,2025)

# create two random arrays for your book dates
book1 = np.array(np.random.randint(low=1950, high=1970, size=50))
book2 = np.array(np.random.randint(low=1950, high=1970, size=50))

# add some athor names
author_names = [f'Author_{x+1}' for x in range(50)]

# Data to a barh graph (sideways bar)
axs.barh(range(len(begin)), end-begin, left=begin, zorder=2, 
color='#007acc', alpha=0.8, linewidth=5)

# Plots the books in a scatterplot. Changes marker color and shape.
axs.scatter(book1, range(len(begin)), color='purple', s=30, marker='D', zorder=3, label='1st Book')

# second array of books
axs.scatter(book2, range(len(begin)), color='yellow', s=30, marker='D', zorder=3, label='2nd Book')

# or plot a custom array of books
# you could do this in a for loop for all authors
axs.scatter(x=[1980,2005], y=[10,45], color='red', s=50, marker='X', zorder=3, label='3rd Book')

# Sets the titles of the y-axis.
axs.set_yticks(range(len(begin)))
axs.set_yticklabels(author_names)

# Add legend
axs.legend()

# Sets start and end of the x-axis.
axs.set_xlim([1895, 2025])
axs.set_ylim([-1,50]);

Answer 3

(Next time please include a dataframe example !)

I would use the great numpy.unique method to perform the grouping operation.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt


dataset = pd.DataFrame({'BORN': [1900, 1920, 1900],
                        'DIED': [1980, 1978, 1980],
                        'AUTHOR': ['foo', 'bar', 'foo'],
                        'YEAR (BOOK)': [1950, 1972, 1961]})

# --group by author
unique_authors, index, reverse_index = np.unique(dataset.AUTHOR.values, return_index=True, return_inverse=True)
authors_df = dataset.loc[index, ['AUTHOR', 'BORN', 'DIED']]
dataset['AUTHOR_IDX'] = reverse_index  # remember the index

# dataframe columns to arrays.
begin = authors_df.BORN.values
end = authors_df.DIED.values
authors = authors_df.AUTHOR.values

# --Author data to a barh graph (sideways bar)
plt.barh(range(len(begin)), end-begin, left=begin, zorder=2, color='#007acc', alpha=0.8, linewidth=5)

# Sets the titles of the y-axis.
plt.yticks(range(len(begin)), authors)

# Sets start and end of the x-axis.
plt.xlim([1835, 2019])

# --Overlay book information
# dataframe columns to arrays
book = dataset['YEAR (BOOK)'].values

# Plots the books in a scatterplot. Changes marker color and shape.
plt.scatter(book, reverse_index, color='purple', s=30, marker='D', zorder=3)

# Shows the plt
plt.show()

Yields: