How adiatgithub/automated_da works
ADA is a Streamlit business-analytics application: users upload tabular files, inspect cleaned and role-detected data, then ask questions or explore generated business insights, charts, anomalies, and forecasts. Its documented core is deterministic and local, with SQL/Pandas translations and reasoning traces; optional AI narrative generation is a separate capability. The README documents the broad product flow, while sampled source confirms preparation, query, aggregation, and presentation responsibilities; some UI wiring and unsampled feature links are therefore left out.
Main entry points: app.py Streamlit app · pipeline.py Analysis preparation · agent.py Analyst agent · aggregation.py Trend aggregation
Components 23
Data preparation
- File ingestion
file_io.py - Analysis preparation
pipeline.py - Data cleaning
analysis.py - Data quality
data_quality.py - Role detection
schema.py
Analysis and reasoning
- Analyst agent
agent.py - Natural-language queries
nlq.py - SQL engine
sql_engine.py - Trend aggregation
aggregation.py - Metric semantics
metrics.py - Time-series methods
timeseries.py - Record search
semantic_search.py - Query evaluation
eval_framework.py
Business intelligence
- Anomaly detection
anomalies.py - Forecasting
forecasting.py - Business insights
business_insights.py - AI narrative
ai_insights.py
Application experience
- Streamlit app
app.py - Chart generation
autovis.py - Streamlit interface
ui.py
Governance
- Role permissions
auth.py - Audit telemetry
observability.py
Other components
- Business user
Connections 26
- Business user → Streamlit app: uses
- Streamlit app → File ingestion: loads files
- File ingestion → Analysis preparation: provides frames
- Analysis preparation → Data cleaning: cleans data
- Analysis preparation → Role detection: detects roles
- Analysis preparation → Trend aggregation: selects periods
- Analysis preparation → Business insights: builds brief
- File ingestion → Data quality: audits data
- Streamlit app → Streamlit interface: renders
- Streamlit app → Analysis preparation: prepares analysis
- Streamlit app → Analyst agent: routes requests
- Analyst agent → Natural-language queries: dispatches queries
- Analyst agent → Anomaly detection: dispatches detection
- Analyst agent → Forecasting: dispatches forecasts
- Analyst agent → Record search: dispatches search
- Natural-language queries → SQL engine: generates SQL
- Natural-language queries → Trend aggregation: builds trends
- Trend aggregation → Metric semantics: resolves metrics
- Trend aggregation → Time-series methods: fits trends
- Business insights → Trend aggregation: analyzes trends
- Business insights → Anomaly detection: checks anomalies
- Streamlit interface → Chart generation: renders charts
- Streamlit interface → Role permissions: checks permissions
- Streamlit interface → Audit telemetry: shows telemetry
- Streamlit app → AI narrative: generates narrative
- Streamlit app → Query evaluation: runs benchmarks