UAI Member Spotlight: Alejandro Komai, SCE
Dr. Alejandro Komai is a data scientist with Maintenance and Inspection Analytics at Southern California Edison. In his current role, he applies machine learning algorithms to predictive maintenance of poles, both pole loading and intrusive inspection. He earned his doctorate in economics from the University of California, Irvine.
1. What types of benefits and ROI has your utility observed from adopting analytics?
Within Southern California Edison, Maintenance, Performance, and Reliability has been working on predictive analytics models — models that help predict which equipment are more likely to fail. The gains we have seen have been on the reliability side; removing assets with high risk of failure reduces the frequency of unplanned outages. We have also experimented with using machine learning to monetize risk.
2. What types of tools are you using for analytics?
Personally, I’m a fan of R. On a daily basis I use R in RStudio. Most of my machine learning work makes use of the caret library. For data munging I use dplyr. My visualizations tend to make use of ggplot2 or Google Earth Pro. I’m learning how to use keras for deep learning right now. There are others at Edison who prefer to use Python or SAS for analytics.
3. What does the future look like at your utility with respect to analytics?
Our director, Joe Goizueta, has committed to making our team a world class data analytics shop. As a part of this commitment, Matthew Mendoza, Sophie Lellis-Petrie, Mark Turgeon, and I are being sent to represent Edison at the SAP-ESRI Hackathon where we aim to make use of deep learning methods to predict and visualize outages. We envision using neural networks to automate identification of equipment in maintenance and inspection images. As our team, including Eric X. Wang, will discuss at the next Utility Analytics Summit in Irvine, using machine learning to build goals for performance enhancement is already beginning to see success. Furthermore, we plan to serve our internal business partners with automated SAS Visual Analytics reports.
4. What working groups do you and other members participate in at UAI?
At the moment we participate in the Asset Optimization and Performance Management Working Group, but we would love to learn more about all the ways analytics is being adopted in the utilities industry.
