Description

Excavation damage are a major concern for utility companies since it lead to:

• Power outages.
• Gas leakages (and safety issues).

Restoring these damages is time-consuming and leads to high costs for society. Therefore, it is desired to accurately prevent those damages in advance.

I worked on:

• Extracting, loading and transforming data such as time, place and polygon of excavation work.
• Combining this with internal utility asset data based on spatial geometries.
• Engineering of features such as complexity of excavation, number of assets in excavation area and the resonsible excavator
• Predicting the probability of excavation damage by a trained Machine learning model.

The end result is an automatically scheduled pipeline, on a daily basis that generate predictions for the outdoor prevention team.

Responsibilities

Project lead and stakeholder management.

Tools

Python, Pandas, scikit-learn, XGBoost, Oracle Geometries, AWS ECS, Docker, Gitlab CI/CD, Airflow.