Try it yourself

Run this code directly in your browser. Click "Open in full editor" to experiment further.

Loading...

Click Run to see output

Or press Ctrl + Enter

How it works

Word clouds turn a wall of text into an instant visual summary — the more often a word appears, the bigger it gets. ☁️

What's a Word Cloud Good For?

Quick text summaries — get the gist of a long article, review batch, or transcript at a glance

Social media analysis — see what topics dominate a hashtag or comment section

Survey responses — visualize what customers actually said in open-ended answers

Blog posts and presentations — they look great as a hero image

NLP exploration — a fast first look at your text data before deeper analysis

How It Works (Behind the Scenes)

1. Tokenize — split your text into individual words.

2. Filter — remove "stopwords" (the, and, is, of...) that would dominate but mean nothing.

3. Count — tally how many times each word appears.

4. Size — bigger count → bigger font.

5. Pack — fit all the words into the image without overlapping, prioritizing the biggest ones.

6. Color — apply a colormap so the result looks visually pleasing.

The wordcloud library handles every step — you just feed it text.

The Three Ways to Make One

1. From raw text — the simplest case

WordCloud(width=800, height=400).generate("your big string of text...")

2. From a frequency dictionary — when you already counted

freqs = {'python': 100, 'javascript': 85, 'rust': 45}
WordCloud().generate_from_frequencies(freqs)

This is the magic method when you've already used pandas.Series.value_counts() or collections.Counter. Skip the text-parsing step entirely.

3. From a file

with open('article.txt') as f:
    text = f.read()
WordCloud().generate(text)

Key Parameters You'll Actually Use

Parameter	What it does
`width`, `height`	Output image size in pixels
`background_color`	`'white'`, `'black'`, or any color name
`colormap`	Any matplotlib colormap: `'viridis'`, `'plasma'`, `'inferno'`, `'cool'`, `'autumn'`
`max_words`	Cap the number of words shown (default: 200)
`stopwords`	A set of words to ignore — combine `STOPWORDS` with your own additions
`min_font_size` / `max_font_size`	Size range for the words
`relative_scaling`	`0` = sizes ignore frequency, `1` = pure frequency-based
`random_state`	Lock the layout so it's reproducible
`mask`	A numpy array image to shape the cloud (heart, country, logo, etc.)

Stopwords — The Most Important Setting

Without stopwords, your cloud will be dominated by the, and, of, a. Always filter them:

from wordcloud import STOPWORDS

stopwords = set(STOPWORDS)             # 200+ common English words
stopwords.update(['said', 'would'])    # add your own domain-specific noise

Displaying the Cloud

A WordCloud object isn't an image yet — it's a data structure. Hand it to matplotlib to render:

plt.imshow(cloud, interpolation='bilinear')   # 'bilinear' = smooth edges
plt.axis('off')                               # hide ticks and labels
plt.show()

Use interpolation='bilinear' (or 'lanczos') — without it, the words look pixelated.

Pro Tips

Start small: build a cloud with 50–100 words before scaling up — easier to read.

Custom shapes: pass a black-and-white image array as mask= to make the cloud fit a logo or country outline.

Save to file: cloud.to_file('cloud.png') exports a PNG directly.

Multilingual: works fine for any language, but stopword lists are English-only by default — add your own for French, Arabic, etc.

Avoid the cliché: word clouds work best for exploration, not for serious analytics. For real frequency comparison, a bar chart is more honest.

Run the snippet above to see three different word clouds: a basic one, a styled dark-theme one, and one built from a frequency dictionary of programming languages. Try swapping in your own text! ✨

Related examples

Data Science

Word Cloud in Python