viz

Concept Overview

The viz module produces structured chart payloads — plain Python dicts with chart type, axis data, and optional error bars or color channels. These payloads are plotting-library-agnostic: you can render them with Plotly, matplotlib, or pass them to a web frontend.

This decouples analysis from visualization: the feature_diagnostics module computes importance scores and calls viz internally to produce payloads, which you can render however you prefer. The pattern keeps the core modules free of plotting dependencies.

When to Use

Use viz payloads when you want structured chart data from research outputs. Most diagnostic modules (feature_diagnostics, pipeline) already call viz internally and include payloads in their return dicts.

Alternatives: Build charts directly from DataFrames if you prefer a specific plotting library’s API.

Usage Examples

Python

Build visualization payloads for research output

from openquant.viz import (
    prepare_feature_importance_payload,
    prepare_drawdown_payload,
)

# Feature importance bar chart payload
payload = prepare_feature_importance_payload(
    feature_names=["momentum", "vol", "spread"],
    importance=[0.45, 0.35, 0.20],
    std=[0.05, 0.03, 0.02],
    top_n=10,
)
# {"chart": "bar", "x": [...], "y": [...], "error_y": [...]}

# Drawdown chart payload from equity curve
dd_payload = prepare_drawdown_payload(
    timestamps=["2024-01-02", "2024-01-03", "2024-01-04"],
    equity_curve=[1.0, 1.02, 0.98],
)
# {"chart": "line", "x": [...], "equity": [...], "drawdown": [...]}

API Reference

Python API

viz.prepare_feature_importance_payload
viz.prepare_feature_importance_comparison_payload
viz.prepare_drawdown_payload
viz.prepare_regime_payload
viz.prepare_frontier_payload
viz.prepare_cluster_payload

Key Functions

prepare_feature_importance_payload
prepare_drawdown_payload
prepare_regime_payload
prepare_frontier_payload
prepare_cluster_payload

Implementation Notes

Payloads are plain dicts — render with plotly, matplotlib, or pass to a frontend.
prepare_feature_importance_payload sorts by importance descending and supports top_n filtering.
prepare_feature_importance_comparison_payload creates side-by-side grouped bar payloads for before/after analysis.

viz