Accurate geolocation data is critical for energy operators, yet often difficult to verify at scale. By applying AI based data refinement techniques, this gas distribution company significantly improved the reliability of its PDR geolocation data, creating a solid foundation for analytics and operational efficiency.

The client is one of the leading natural gas distribution operators in Italy, managing a large number of Points of Delivery (PDRs) across the national territory and supporting complex operational and analytical processes.

The geolocation of PDRs presented significant data quality issues.

In particular:

  • Large volumes of geolocation data were available, but difficult to verify
  • Data quality was inconsistent and unsuitable as a baseline for machine learning models
  • Geolocation data needed to be combined with other datasets, increasing complexity
  • There was no mechanism to validate future incoming data

Without reliable geolocation information, both operational efficiency and analytical accuracy were compromised.

The solution is based on AI and data science techniques designed to refine and validate geospatial data.

Machine learning models analyze existing coordinates, cross referencing them with external and contextual data sources to assess reliability and apply corrections where possible. The system is trained on historical data and continuously improves over time.

BitBang designed and implemented an AI powered data refinement system capable of:

  • Evaluating the reliability of existing PDR geolocation data
  • Correcting inaccurate coordinates when possible
  • Iteratively improving data quality, starting from the most verifiable records
  • Applying the same validation logic to newly acquired data

A dedicated dashboard was also developed to monitor system performance and support data quality control activities.

The project delivered tangible results:

  • 77% of PDR geolocation records were validated or corrected and are now considered reliable
  • Improved efficiency in operational and analytical processes
  • Refined data was successfully integrated into existing systems
  • Increased accuracy and consistency across data driven analyses

The organization now benefits from a continuously improving, future proof geolocation dataset.