如何使用Bokeh读取JSON文件并基于Pandas DataFrame创建环形饼图

Let's walk through the corrections and improvements to get your donut chart working properly:

Key Issues in Your Original Code #

• Typo: You imported donut (lowercase) but used Donut (capitalized) when creating the chart.
• Missing Value Column: Your JSON data only has categorical fields (name, task), but the donut chart requires numerical values (like counts of each task/person pair).
• Data Reading: Using Pandas' built-in read_json is more reliable than Bokeh's df_from_json.

Corrected Step-by-Step Code #

1. Import Required Libraries #

First, fix the import for Donut (match the capitalization):

from bokeh.charts import Donut, show, output_file
import pandas as pd

2. Read JSON Data into Pandas DataFrame #

Use Pandas to load your JSON file:

# Replace with your actual file path
df = pd.read_json('path/to/your/data.json')

3. Aggregate Data to Get Counts #

Since your JSON doesn't include numerical values, we need to count how many times each (name, task) pair appears:

# Group by name and task, then calculate counts
df_counts = df.groupby(['name', 'task']).size().reset_index(name='count')

4. Create the Donut Chart #

Now use the aggregated counts to build your nested donut (outer ring = names, inner ring = tasks):

# Initialize donut chart with correct parameters
d = Donut(
    df_counts,
    label=['name', 'task'],  # Nested labels for outer/inner rings
    values='count',          # Use our calculated count values
    text_font_size='8pt'
)

5. Output and Display the Chart #

output_file("donut.html")
show(d)

Example Output Explanation #

If your JSON data looks like this:

[
  {"name": "Anand", "task": "development"},
  {"name": "Anand", "task": "testing"},
  {"name": "Priya", "task": "development"},
  {"name": "Priya", "task": "design"},
  {"name": "Priya", "task": "testing"}
]

The resulting donut will:

• Have outer segments for Anand (2 tasks) and Priya (3 tasks)
• Split each outer segment into inner segments for their specific tasks (e.g., Anand's segment splits into development and testing, each with count 1)

Notes #

• Ensure your file path is correct (use absolute path if relative paths aren't working)
• If you're using a newer version of Bokeh (post-0.13), the bokeh.charts module is deprecated—you'll need to use bokeh.plotting with wedge glyphs instead, but the above code works for older versions where bokeh.charts is available.

内容的提问来源于stack exchange，提问作者Nightwing