fix(scatterplot): address an issue where maidr.show rendering is sluggish when there are massive data points

[jooyoungseo]

Reproducible Steps

1. Execute the following code in the Python repl:

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import maidr
diamonds = sns.load_dataset("diamonds")

# Scatter plot for Carat vs price in the diamonds dataset
plt.figure(figsize=(10, 6))
karrat = sns.scatterplot(data=diamonds, x="carat", y="price")
plt.title("Carat vs Price")
plt.xlabel("Carat")
plt.ylabel("Price")
plt.show()
maidr.show(karrat)

1. Tab to the interactive maidr plot area in the browser and use RightArrow to navigate the data points.

Current Issue

maidr does not respond.

[SaaiVenkat]

Professor @jooyoungseo,

Benchmark Timings

• plt.savefig(): 450 milliseconds (on average)
• maidr.save_html(): 710 milliseconds (on average)

Observation

• Even though the increase in time seems significant (around 50%), I believe the overall processing time is well within acceptable range.
• The more time corresponds to the extraction of data and the injection of custom property for highlighting.

Thoughts

• Python binder performs only the extraction, and the rendering is performed by the upstream library.
• Since the datapoints are so many, handling the corresponding highlighting and sonification events makes the rendering sluggish. We should investigate more with the JS library.

[SaaiVenkat]

Moved the issue to the upstream repo - xability/maidr#518