Using Machine Learning to Predict Building Fire Risk

Current Projects: Metro21 Fire Risk

Working with the Pittsburgh Bureau of Fire (PBF), the Fire Risk Analysis team is using historical fire incident and inspection data, coupled with business permits and property condition information to develop predictive models of structure fire risk for commercial properties. PBF conducts regular fire inspections of commercial properties, as stipulated by the municipal fire code. With a prioritization method informed by the prediction probability of fire risk from machine learning models, and implemented in an interactive map visualization, they will be better able to target their inspections for the properties at greatest risk of fire.

You can see more about this project at our project page here. After deploying the model on the City of Pittsburgh's servers for usage by city officials, we released a technical report describing our process in detail, so that fire departments from other cities could adopt our approach and use our code (available on GitHub here).

This work was highlighted by the GovTech and MetroLab Network as their "Innovation of the Month" in January, 2018.

In April, 2018, Pittsburgh's Mayor Bill Peduto anounced the city's use of machine learning to improve their fire risk reduction practices in a press conference.

We published a paper in the 2018 Knowledge Discovery and Data Mining (KDD) conference describing our spatio-temporal modeling process and the evaluation of our model's performance over time for predicting both residential and commercial property fire risk:

Singh, B., Hu, Q., Chen, J., Chen, F., Lee, J., Kuo, N., Narang, P., Batts, J., Arnold, G., Madaio, M. (2018). A dynamic pipeline for spatio-temporal fire risk prediction. In Proceedings of the 2018 ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) (in press).