Case Study: Technical Safety BC achieves an 85% improvement in predictive performance and optimizes inspection resource allocation with Dataiku

Real-Time Predictions at Technical Safety BC for Targeted Safety Oversight

Technical Safety BC, an independent, self-funded safety regulator in British Columbia, needed to find more high-hazard sites without increasing inspection resources. Faced with uncoordinated heterogeneous data, data-quality and training challenges, they built a decision-science team and selected Dataiku as the platform to introduce machine-learning-driven risk assessment and enable rapid prototyping, model reuse, and collaborative development.

Using Dataiku, the team quickly iterated, A/B tested, and deployed adaptive predictive models that automatically reprioritize inspections. Dataiku-powered models improved predictive performance by 85% for electrical technology versus previous methods, substantially reduced the total number of mandatory inspections, optimized safety-officer time toward higher-risk sites, and delivered faster, more scalable deployments and greater organizational trust in machine learning.

Technical Safety BC

Soyean Kim

Leader, Research and Analytics