Scatter Plots & Regression

Purpose

Scatter plots reveal relationships between two continuous variables. Adding a regression line or smoothed curve helps identify trends and fit patterns.

Example in the Notebook

The notebook uses the iris dataset to plot:

1. Sepal length vs. sepal width as a scatter plot, colored and styled by species.
2. A LOWESS (Locally Weighted Scatterplot Smoothing) regression curve overlay.

Key Code Snippet

sns.scatterplot(data=iris, x='sepal_length', y='sepal_width', hue='species', style='species', ax=ax, s=60)
sns.regplot(data=iris, x='sepal_length', y='sepal_width', scatter=False, ax=ax, 
            color='gray', lowess=True)

Customization Tips

• Point size: Change s=60 to adjust marker size.
• Regression method: Use lowess=True for non-parametric smoothing or omit for linear regression.
• Color palettes: Customize the hue palette with palette='Set2' or others.
• Transparency: Add alpha=0.6 to scatter points for overlapping data visibility.
• Faceting: Use sns.relplot(..., col='species') to create subplots per category.

When to Use

• Always: for exploring correlations between two continuous variables.
• Consider: adding transparency or jitter when points overlap heavily.
• Avoid: when you have too many points (>10,000) without aggregation; use hexbin or datashader instead.

See Also

• Seaborn: scatterplot(), regplot()
• Matplotlib: scatter()