src.llm_interpreter.analysis.insights

Insights combiner for AMMM analysis.

This module combines quantitative summaries and rule-based diagnostics into structured insights ready for LLM processing or direct reporting.

Author: AMMM Team Created: 2025-04-10 Last Modified: 2025-04-10

Module Contents

class src.llm_interpreter.analysis.insights.StructuredInsights

Structured insights combining summaries and diagnostics.

This is the primary output of the analysis engine, providing evidence-backed insights ready for LLM narrative generation.

classmethod from_csv_data(csv_data: dict[str, list]) → StructuredInsights

Create structured insights from loaded CSV data.

Parameters:

csv_data – Dictionary mapping CSV names to lists of dataclass instances

Returns:

StructuredInsights with all analysis results

get_executive_summary_data() → dict[str, Any]

Extract key data points for executive summary.

Returns:

Dictionary with executive summary data points

get_data_quality_insights() → dict[str, Any]

Extract data quality insights.

Returns:

Dictionary with data quality metrics and issues

get_model_fit_insights() → dict[str, Any]

Extract model fit insights.

Returns:

Dictionary with model fit metrics and issues

get_media_insights() → dict[str, Any]

Extract media performance insights.

Returns:

Dictionary with media performance data

get_budget_optimization_insights() → dict[str, Any]

Extract budget optimization insights.

Returns:

Dictionary with budget scenario analysis

to_dict() → dict[str, Any]

Convert insights to dictionary for JSON serialization or LLM processing.

Returns:

Complete insights as dictionary

src.llm_interpreter.analysis.insights.analyze_csv_data(csv_data: dict[str, list]) → StructuredInsights

Main entry point for analyzing CSV data.

This is the primary function to call from external code.

Parameters:

csv_data – Dictionary mapping CSV names to lists of dataclass instances

Returns:

StructuredInsights with complete analysis

Example

>>> from src.llm_interpreter.inputs import load_all_csvs
>>> from src.llm_interpreter.analysis import analyze_csv_data
>>>
>>> csv_data = load_all_csvs("results/csv")
>>> insights = analyze_csv_data(csv_data)
>>>
>>> print(f"Found {len(insights.critical_findings)} critical issues")
>>> print(f"Top channel: {insights.summary.media_performance.top_performer}")