Pulp Mill Effluent Compliance: An Engineering-First Data Pipeline

📊 Business Intelligence & Industrial Compliance

This project transforms raw effluent sensor data from a pulp mill into an interactive regulatory dashboard. It bridges the gap between process engineering and data science by implementing industrial standards directly into the ETL pipeline.

🎯 Key Performance Indicators (KPIs)

• Overall Compliance Rate: 36.7%
• Financial Risk: Calculated based on a $500/violation model.
• Primary Violation Driver: Turbidity (averaging 4.45 NTU during non-compliance).

Data source for case study: https://www.kaggle.com/datasets/adityakadiwal/water-potability

---

🛠️ Data Engineering & "Data Dignity"

The pipeline was built in Power Query with a focus on data integrity:

1. Locale-Aware Ingestion: Applied en-US locale settings to ensure decimal precision (protecting pH and Conductivity readings).
2. Dynamic Mean Imputation: Automated null-handling for 491 missing sensor records using a scalable M-code script (List.Accumulate).
3. Biological Sanity Checks: Verified all pH data falls within the 0.0–14.0 range to eliminate sensor artifacts.

---

📂 Project Structure

• /data_raw: Original sensor dataset.
• /data_processed: Final Excel dashboard featuring DAX measures and Root Cause Heatmaps.
• /scripts: Full M-code documentation for the ETL process.
• /docs: Audit logs, terminology, and industrial standards used for the model.

---

🔗 Links

• Medium Article: [https://medium.com/@andersonrumuy/protecting-data-dignity-an-engineering-first-approach-to-pulp-mill-analytics-9dcaa3a587f0]
• LinkedIn Profile: [https://www.linkedin.com/in/andersonrumuy/]