Plot to See Patterns

pandas borrows matplotlib, so plotting is one method call:

• df['price'].plot.hist(bins=50) — a histogram reveals skew, outliers, and the long tail describe() hides
• df.plot.scatter(x='sqft', y='price') — relationships, clusters, and the point that's 100× too big
• plt.plot(epochs, losses) — the training curve, ML's favorite chart

For prettier defaults and group coloring, seaborn wraps the same engine: sns.scatterplot(data=df, x='sqft', y='price', hue='neighborhood') colors points by category with a single argument. A figure that takes 30 seconds routinely saves three days of modeling the wrong thing.