Treemap Visualisations

Treemaps turn the attribution output from interpret() into an interactive HTML chart. The chart is hierarchical — data sources, features, feature values — so you can zoom in from a high-level overview to individual value contributions.

Default Treemap

Point TreemapGenerator at the interpretation output directory and call plot_treemap():

from pathlib import Path
from monad.interpretability.treemap import TreemapGenerator

generator = TreemapGenerator(
    interpretability_files_path=Path("./interpretations"),
)

generator.plot_treemap(
    output_file_path=Path("./treemap.html"),
)

Open the saved HTML file in a browser — it is fully interactive (hover, zoom, pan).

Reading the treemap

The top level shows feature importance within each data source. Area encodes the magnitude of the attribution; color encodes the direction (green = positive, red = negative).

Click a feature to drill into individual value attributions. Here the price feature is expanded to show quantile-range contributions:

For categorical features the picture is simpler — each value gets its own tile with a clear positive or negative attribution:

Performance Tuning

Large datasets with many feature values can make the chart slow to render. Use these parameters to reduce the volume:

• n_largest_per_feature — maximum values shown per feature (default 1500)
• n_largest — maximum total values across all features
• max_depth — hierarchy depth rendered at once (default 3; set to -1 for all levels)

generator.plot_treemap(
    output_file_path=Path("./treemap.html"),
    n_largest_per_feature=500,
    max_depth=2,
)

Custom Hierarchy

The default treemap groups by data source, then feature, then value. When you need a domain-specific grouping (e.g., product category / brand / item), define a TreemapHierarchy:

from monad.interpretability.treemap import TreemapGenerator, TreemapHierarchy

hierarchy = TreemapHierarchy(
    levels=["category", "brand", "product_id"],
    hierarchy_path=Path("./data/product_hierarchy.csv"),
    feature_values_importance_path=Path(
        "./interpretations/transactions/article_id/values_importance.json"
    ),
    entity_name_column="product_name",
)

generator = TreemapGenerator(hierarchy=hierarchy)
generator.plot_treemap(output_file_path=Path("./custom_treemap.html"))

The CSV at hierarchy_path maps leaf-level values to their parent groupings. See Reference: Interpretability for field details.