Plotly Express Fundamentals

From Static to Interactive: Why Plotly?

Matplotlib is the workhorse of Python plotting, but it was designed for creating static, publication-quality figures. Plotly, and specifically its high-level API, Plotly Express, is designed from the ground up for interactivity in web-based environments like Jupyter notebooks.

With Plotly, you can:

• Hover over data points to see their exact values.
• Zoom and pan to inspect dense areas of your plots.
• Click on legend items to toggle data series on and off.
• Create animations to see how data evolves over time.

These features make it an invaluable tool for exploratory data analysis (EDA). To get a better sense of Plotly and its place in the Python visualization ecosystem, let's start with a brief reading.

The Plotly Express Workflow

As you just read, Plotly Express (px) is a high-level interface that makes creating beautiful, interactive plots incredibly simple. The syntax is very similar to the pandas plotting functions you saw in the last lesson and is designed to work seamlessly with pandas DataFrames.

The general pattern is straightforward:

1. Import the library: import plotly.express as px
2. Import pandas to handle your data: import pandas as pd
3. Load your data into a DataFrame.
4. Call the appropriate px function (e.g., px.scatter, px.bar) with your DataFrame and column names.
5. Call fig.show() to display the interactive plot.

A Tour of Interactive Plots with Plotly Express

Now for the fun part. Let's create the interactive versions of the plots we learned about in the previous lesson. We'll use a comprehensive video tutorial to demonstrate the creation and interactivity of each plot type. I will provide code summaries for you to follow along.

For our examples, we will use some of the built-in datasets that come with Plotly Express, such as the tips dataset. You can load it easily:
df_tips = px.data.tips()

1. Scatter Plots (px.scatter)

Just like in Matplotlib, scatter plots are used to visualize the relationship between two numerical variables. With Plotly, you can hover over each point to see the underlying data.

Let's see how to create one and add a third dimension of information using color.

Code Summary (using the tips dataset):

import plotly.express as px




# Load the built-in tips dataset
df_tips = px.data.tips()




# Create an interactive scatter plot
fig = px.scatter(df_tips, 
                 x="total_bill", 
                 y="tip", 
                 color="smoker",      # Color points by 'smoker' column
                 size='size',       # Size points by the 'size' of the party
                 hover_data=['day'],  # Show the 'day' column on hover
                 title="Tip Amount vs. Total Bill")

fig.show()

Run this code yourself. Hover over the points and click the "smoker" and "non-smoker" items in the legend to see the interactivity in action!

2. Bar Charts (px.bar)

Bar charts are for comparing quantities across different categories. Plotly makes it easy to create simple, stacked, or grouped bar charts.

Code Summary (Grouped Bar Chart):
To create a grouped bar chart (side-by-side) instead of a stacked one, you simply add the barmode='group' argument.

import plotly.express as px
df_tips = px.data.tips()




# Create a grouped bar chart
fig = px.bar(df_tips, 
             x="day", 
             y="total_bill", 
             color="smoker",
             barmode="group", # This creates side-by-side bars
             title="Total Bill by Day, Grouped by Smoker")

fig.show()

3. Histograms (px.histogram)

Histograms show the distribution of a single numerical variable. Plotly's histograms are interactive, allowing you to see the range and count for each bin on hover.

Code Summary:

import plotly.express as px
df_tips = px.data.tips()




# Create a histogram of total_bill, with a box plot in the margin
fig = px.histogram(df_tips, 
                   x="total_bill", 
                   color="day",       # Create overlapping histograms for each day
                   marginal="box",    # Add a box plot to the top margin
                   title="Distribution of Total Bill Amounts by Day")

fig.show()

4. Box Plots (px.box) and Violin Plots (px.violin)

While histograms are great, box plots and violin plots are often better for comparing distributions across several categories.

• Box plots show the quartiles, median, and outliers.
• Violin plots combine a box plot with a density curve, showing the shape of the distribution.

Code Summary (Violin Plot with inner Box Plot):

import plotly.express as px
df_tips = px.data.tips()




# Create a violin plot to compare tip distributions by day
fig = px.violin(df_tips, 
                x="day", 
                y="tip", 
                color="sex",
                box=True,       # Display a box plot inside the violin
                points="all",   # Show all the underlying data points
                title="Tip Distribution by Day and Sex")

fig.show()

Advanced Interactivity

Plotly Express makes it surprisingly easy to create very advanced visualizations. Let's look at two powerful features: faceting and animation.

• Faceting: Creating a grid of subplots based on a category. This is done with the facet_col and/or facet_row parameters.
• Animation: Creating an animation that shows how the data changes based on a variable (like time). This is done with the animation_frame parameter.

Saving and Sharing Your Plots

Once you've created a plot, you'll often want to save it. You have two main options:

1. Interactive HTML file: This saves the plot as a self-contained web page, preserving all interactivity.
2. Static image file (PNG, JPEG, SVG, PDF): This is for use in reports or presentations where interactivity isn't possible.

Conclusion

Congratulations! You've taken a significant step forward in your data visualization capabilities. You've moved beyond static charts and can now build rich, interactive visualizations that make data exploration faster and more intuitive.

For a quick reference, here is a handy cheat sheet that summarizes the key functions in Plotly Express.

Key Takeaways:

• Plotly Express (px) is a high-level Python library for creating interactive visualizations with simple syntax.
• It integrates seamlessly with pandas DataFrames.
• You can create interactive versions of common plots: px.scatter, px.bar, px.histogram, px.line, px.box, and px.violin.
• You can add more data dimensions using parameters like color, size, symbol, and hover_data.
• Advanced features like faceting (facet_col) and animations (animation_frame) can reveal complex patterns.
• Plots can be saved as interactive HTML files (fig.write_html()) or static images (fig.write_image()).

In our next lesson, we will put everything from this module together. You will learn how to perform exploratory data analysis and document key findings. The skills you've gained in data wrangling with pandas and visualization with both Matplotlib and Plotly will be the core tools you use to investigate a dataset from start to finish.