Natural Language Processing Reveals Core Issues in Uganda's Power Grid: A Study of Outages from 2015-2022

This research study investigates the root causes of power outages in Uganda, leveraging a unique data set obtained through a partnership with the country's electricity regulator. Covering the period from 2015 to 2022, this data set includes detailed records of each power outage, noting the duration and free-form descriptions of the incidents. The researchers used an advanced natural language processing (NLP) tool to categorize these descriptions into specific root cause groups. The findings highlight equipment failure as the most prevalent cause of power outages, suggesting a critical need to enhance or reevaluate maintenance efforts. Furthermore, the study revealed that outages resulting from vandalism and fires tend to have significantly longer resolution times, indicating areas where targeted interventions could improve response efficiency. This research sheds light on the main factors contributing to power outages in Uganda and demonstrates how NLP can be applied to quickly and accurately identify these causes. The insights gained from this study offer valuable guidance for policymakers and utility companies aiming to improve the reliability and resilience of Uganda's power grid.