Box Office Explorer

Use inflation-adjusted (real) box office

Metric:

⚠️ Pre-1970 medians rest on very few surviving films — treat as noisy.

Factors to include in the matrix (Ctrl/⌘-click; box-office targets always shown)

Correlation Show values

Pairwise correlation among the selected factors and box office, recomputed on your current selection. $ and skewed counts are log-transformed; pairs use all rows where both values exist.

Each factor's correlation with log₁₀(real box office) — the box-office column of the matrix above, sorted.

Dimensions (Ctrl/⌘-click; 2–6 works best)

Color by

Scatter-plot matrix of the selected factors on your current selection (max 1500 points sampled for speed). Log-scaled factors are labelled “(log)”.

Scenario (precomputed gradient-boosting model):

–

5-fold CV R²

–

Films

–

Features

Model results are precomputed on the full dataset (sklearn can't run in a static page). Pre-release R² is the honest signal — how predictable box office is from what's known before release; "ALL" leaks post-release audience reaction.

Click a column header to sort. Showing current selection. $ values in millions.