Alternatives for big data (Datashader, HoloViews) in Matplotlib - Mini Project: Build & Apply

Visualizing Large Data with Datashader and HoloViews

📖 Scenario: You work as a data analyst for a city planning team. They have collected millions of GPS points from taxis to understand traffic patterns. Plotting all points with normal tools is slow and unclear. You will learn how to use Datashader and HoloViews to visualize big data efficiently.

🎯 Goal: Create a large dataset of GPS points, set a plotting configuration, use Datashader with HoloViews to visualize the data, and display the result.

📋 What You'll Learn

Create a dictionary called taxi_data with keys 'x' and 'y' containing lists of 100000 random float values between 0 and 100

Create a variable called plot_width and set it to 600

Use hv.Points with taxi_data and apply datashade with the width set to plot_width

Print the resulting HoloViews object

💡 Why This Matters

🌍 Real World

City planners and data scientists often work with huge GPS datasets to understand traffic and movement patterns. Visualizing millions of points quickly helps make better decisions.

💼 Career

Skills in Datashader and HoloViews are valuable for data scientists and analysts working with big data visualization, especially in urban planning, transportation, and geospatial analysis.

1

Create the large GPS dataset

Create a dictionary called taxi_data with keys 'x' and 'y'. Each key should map to a list of 100000 random float values between 0 and 100. Use numpy.random.uniform(0, 100, 100000).tolist() for both lists.

Matplotlib

import numpy as np
# Create the taxi_data dictionary with 'x' and 'y' keys containing 100000 random floats between 0 and 100
# Your code here

Need a hint?

Use np.random.uniform(0, 100, 100000).tolist() to generate the lists for both 'x' and 'y'.

2

Set the plot width configuration

Create a variable called plot_width and set it to 600.

Matplotlib

import numpy as np
taxi_data = {
    'x': np.random.uniform(0, 100, 100000).tolist(),
    'y': np.random.uniform(0, 100, 100000).tolist()
}
# Set plot_width to 600
# Your code here

Need a hint?

Just assign the number 600 to the variable plot_width.

3

Create the Datashader plot with HoloViews

Import holoviews as hv. Then import datashade from holoviews.operation.datashader. Initialize HoloViews with hv.extension('bokeh'). Create a hv.Points object called points using taxi_data. Use datashade(points, width=plot_width) to create a shaded plot called shaded_plot.

Matplotlib

import numpy as np
taxi_data = {
    'x': np.random.uniform(0, 100, 100000).tolist(),
    'y': np.random.uniform(0, 100, 100000).tolist()
}
plot_width = 600

# Import holoviews as hv
# from holoviews.operation.datashader import datashade
# Initialize hv extension with 'bokeh'
# Create hv.Points called points from taxi_data
# Use datashade on points with width=plot_width and assign to shaded_plot
# Your code here

Need a hint?

Remember to import datashade from holoviews.operation.datashader and initialize HoloViews with hv.extension('bokeh').

4

Display the shaded plot

Print the shaded_plot object to display the visualization.

Matplotlib

import numpy as np
import holoviews as hv
from holoviews.operation.datashader import datashade

hv.extension('bokeh')

taxi_data = {
    'x': np.random.uniform(0, 100, 100000).tolist(),
    'y': np.random.uniform(0, 100, 100000).tolist()
}
plot_width = 600

points = hv.Points(taxi_data)
shaded_plot = datashade(points, width=plot_width)

# Print shaded_plot to display the result
# Your code here

Need a hint?

Use print(shaded_plot) to show the plot object. The exact output may vary but should include the HoloViews object type.