An end-to-end Neo4j-based manufacturing analytics platform that models production data as a knowledge graph and performs explainable root-cause analysis of quality failures.
Manufacturing quality problems often involve relationships between machines, process parameters, materials, batches and measurements.
Traditional tabular analysis can make these relationships difficult to investigate.
This project models manufacturing processes as a knowledge graph to enable relationship-aware root-cause analysis.
Manufacturing Data ↓ Python ETL Pipeline ↓ Neo4j Knowledge Graph ↓ Cypher Analytics ↓ Python Root-Cause Engine ↓ Streamlit Dashboard ↓ Explainable Graph Visualization
- Python
- Neo4j
- Cypher
- Pandas
- Streamlit
- Graphviz
- pytest
- Git
Example entities:
- Batch
- Product
- ProcessRun
- Machine
- Process
- Material
- Measurement
Relationships capture how manufacturing entities interact during production.
Example:
Batch → ProcessRun → Machine
Batch → ProcessRun → Material
Batch → ProcessRun → Measurement
The RCA engine analyzes evidence related to:
- abnormal process parameters
- machine history
- material lots
- manufacturing relationships
- quality outcomes
Potential causes are ranked using evidence scores.
The scores represent evidence prioritization and should not be interpreted as causal probabilities.
The Streamlit dashboard provides:
- manufacturing batch selection
- quality-status overview
- ranked root-cause evidence
- knowledge-graph visualization
- manufacturing relationship explorer
- detailed RCA output
- JSON report export
- CSV evidence export
Run:
pytest -v