Combine wordcloud and line chart

We will create a wordcloud combined with a line chart using the Matplotlib library to visualize data about movie ratings. The final contains nice annotations, an image on top and cool color styling.

This post is a step-by-step guide on how to create it and provides reproducible code.

About

This plot is a combination of a wordcloud and a line-chart. It shows the evolution of the average rating of different movies since the 1960, with the title of the movies in the background for the wordcloud.

The chart was made by Joseph B.. Thanks to him for accepting sharing his work here!

Let's see what the final picture will look like:

wordcloud and linechart

Libraries

First, you need to load a whole bunch of libraries:

# data
import numpy as np
import pandas as pd
from PIL import Image

# plot
from wordcloud import WordCloud
import matplotlib.pyplot as plt
import matplotlib
import seaborn as sns
from mpl_toolkits.axes_grid1.inset_locator import inset_axes

# annotations
from highlight_text import fig_text, ax_text
from matplotlib.patches import FancyArrowPatch

Dataset

For this reproduction, we're going to retrieve the data directly from the gallery's Github repo. This means we just need to give the right url as an argument to pandas' read_csv() function to retrieve the data.

The dataset contains one row per movie and its rating.

# Open the dataset from Github
url = "https://raw.githubusercontent.com/holtzy/the-python-graph-gallery/master/static/data/movies-review.csv"
df = pd.read_csv(url, usecols=['name', 'year', 'type', 'rating'])
df.head()

	name	year	rating	type
0	The Dark Knight	2008	9.0	Action
1	Pulp Fiction	1994	8.9	Crime
2	The Godfather	1972	9.2	Crime
3	Se7en	1995	8.6	Crime
4	The Silence of the Lambs	1991	8.6	Crime

Text for the wordcloud

Let's start by create the background of the whole chart.

For this we use the wordcloud library. We need to create a string that contains all the titles of the movies. We will use the name column of the dataset for this.

Then, we make it less opaque by setting the alpha parameter to a low value.

# create a large string with all the titles
titles_combined = ' '.join(df['name'])
titles_combined = list(set(titles_combined.split(' ')))
titles_combined = [word.lower() for word in titles_combined if len(word) > 3]
to_remove = ['man', 'world']  # remove some words that are not relevant
titles_combined = [word for word in titles_combined if word not in to_remove]
titles_combined = ' '.join(titles_combined)

# initiate a figure
fig, ax = plt.subplots(1, 1, figsize=(8, 6))

# create the wordcloud for the background
wordcloud = WordCloud(
    width=800, height=600,
    background_color='white',
    colormap='Reds',
    contour_width=1,
    contour_color='black',
    random_state=2  # fix randomness
).generate(titles_combined)
ax.imshow(wordcloud, interpolation='bilinear', alpha=0.07)
ax.set_axis_off()

# display chart
plt.tight_layout()
plt.show()

Line chart on top

For a dedicated post on how to add a chart on top of another, check this post.

# create a large string with all the titles
titles_combined = ' '.join(df['name'])
titles_combined = list(set(titles_combined.split(' ')))
titles_combined = [word.lower() for word in titles_combined if len(word) > 3]
to_remove = ['man', 'world']  # remove some words that are not relevant
titles_combined = [word for word in titles_combined if word not in to_remove]
titles_combined = ' '.join(titles_combined)

# initiate a figure
fig, ax = plt.subplots(1, 1, figsize=(8, 6))

# create the wordcloud for the background
wordcloud = WordCloud(
    width=800, height=600,
    background_color='white',
    colormap='Reds',
    contour_width=1,
    contour_color='black',
    random_state=2  # fix randomness
).generate(titles_combined)
ax.imshow(wordcloud, interpolation='bilinear', alpha=0.07)
ax.set_axis_off()

# create inside axe
sub_ax = inset_axes(
    parent_axes=ax,
    width="70%",
    height="50%",
    loc='center',
    borderpad=1
)

# compute average rating per year
avg_rating = df.groupby('year')['rating'].mean().reset_index()
sns.lineplot(
    data=avg_rating,
    x='year', y='rating',
    color='darkred',
    ax=sub_ax,
    zorder=5
)

# display chart
plt.tight_layout()
plt.show()

/var/folders/m2/jfzxyg4s0xz3jyw5f94r7tcm0000gn/T/ipykernel_29511/3519991070.py:44: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.
  plt.tight_layout()

Add points on the line and remove axis

The main interest of the wordcloud in the background is only to be fully visible. For this, we have to remove the background of the axis on the top.

We also add individual points on the line to make it more readable. Colors of the dots with be proportional to the rating of the movie in order to highlight the pattern.

# create a large string with all the titles
titles_combined = ' '.join(df['name'])
titles_combined = list(set(titles_combined.split(' ')))
titles_combined = [word.lower() for word in titles_combined if len(word) > 3]
to_remove = ['man', 'world']  # remove some words that are not relevant
titles_combined = [word for word in titles_combined if word not in to_remove]
titles_combined = ' '.join(titles_combined)

# initiate a figure
fig, ax = plt.subplots(1, 1, figsize=(8, 6))

# create the wordcloud for the background
wordcloud = WordCloud(
    width=800, height=600,
    background_color='white',
    colormap='Reds',
    contour_width=1,
    contour_color='black',
    random_state=2  # fix randomness
).generate(titles_combined)
ax.imshow(wordcloud, interpolation='bilinear', alpha=0.07)
ax.set_axis_off()

# create inside axe
sub_ax = inset_axes(
    parent_axes=ax,
    width="70%",
    height="50%",
    loc='center',
    borderpad=1
)

# compute average rating per year
avg_rating = df.groupby('year')['rating'].mean().reset_index()
sns.lineplot(
    data=avg_rating,
    x='year', y='rating',
    color='darkred',
    ax=sub_ax,
    zorder=5
)

# add scatter plot on top
cmap = matplotlib.colormaps['Reds']
sub_ax.scatter(
    x=avg_rating['year'],
    y=avg_rating['rating'],
    cmap=cmap,
    c=avg_rating['rating'],
    s=70,
    edgecolor='black',
    linewidth=0.5,
    zorder=10,
)

# remove background from line chart
sub_ax.set_axis_off()

# display chart
plt.tight_layout()
plt.show()

/var/folders/m2/jfzxyg4s0xz3jyw5f94r7tcm0000gn/T/ipykernel_29511/956014063.py:60: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.
  plt.tight_layout()

Scale and reference values

The chart is starting to look really nice, but it lacks scale and context. We can add a reference lines and text to give more information to the reader.

# create a large string with all the titles
titles_combined = ' '.join(df['name'])
titles_combined = list(set(titles_combined.split(' ')))
titles_combined = [word.lower() for word in titles_combined if len(word) > 3]
to_remove = ['man', 'world']  # remove some words that are not relevant
titles_combined = [word for word in titles_combined if word not in to_remove]
titles_combined = ' '.join(titles_combined)

# initiate a figure
fig, ax = plt.subplots(1, 1, figsize=(8, 6))

# create the wordcloud for the background
wordcloud = WordCloud(
    width=800, height=600,
    background_color='white',
    colormap='Reds',
    contour_width=1,
    contour_color='black',
    random_state=2  # fix randomness
).generate(titles_combined)
ax.imshow(wordcloud, interpolation='bilinear', alpha=0.07)
ax.set_axis_off()

# create inside axe
sub_ax = inset_axes(
    parent_axes=ax,
    width="70%",
    height="50%",
    loc='center',
    borderpad=1
)

# compute average rating per year
avg_rating = df.groupby('year')['rating'].mean().reset_index()
sns.lineplot(
    data=avg_rating,
    x='year', y='rating',
    color='darkred',
    ax=sub_ax,
    zorder=5
)

# add scatter plot on top
cmap = matplotlib.colormaps['Reds']
sub_ax.scatter(
    x=avg_rating['year'],
    y=avg_rating['rating'],
    cmap=cmap,
    c=avg_rating['rating'],
    s=70,
    edgecolor='black',
    linewidth=0.5,
    zorder=10,
)

# remove background from line chart
sub_ax.set_axis_off()

# display references
max_rate = 9
min_rate = 7
mid_rate = 8
sub_ax.axhline(y=max_rate, color='lightgrey', linestyle='--', lw=0.6)
sub_ax.axhline(y=min_rate, color='lightgrey', linestyle='--', lw=0.6)
sub_ax.axhline(y=mid_rate, color='lightgrey', linestyle='--', lw=0.6)
sub_ax.text(
    1950, max_rate, f"{max_rate:.0f}/10",
    fontsize=10, color='darkgrey'
)
sub_ax.text(
    1950, min_rate, f"{min_rate:.0f}/10",
    fontsize=10, color='darkgrey'
)
sub_ax.text(
    1950, mid_rate, f"{mid_rate:.0f}/10",
    fontsize=10, color='darkgrey'
)

# vertical line arrow for y axis of the subax
ax.arrow(
    0.13, 0.24, 0, 0.53,
    head_width=0.02, head_length=0.02,
    fc='black', ec='black',
    transform=ax.transAxes
)

# display chart
plt.tight_layout()
plt.show()

/var/folders/m2/jfzxyg4s0xz3jyw5f94r7tcm0000gn/T/ipykernel_29511/3469529372.py:88: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect.
  plt.tight_layout()