Improving Comcast’s Outage Detection System
Student Team
Ian Balmaseda, Thien Huynh, Daniel Kohlsdorf, Sagar Patel, Alejandro Santelises, Holly Thomasson, and Michelle Wang
Faculty Advisor
Alexander Shapiro
The project focused on reducing the number of unnecessary trucks and service calls that Comcast incurred because of inaccurate outage detection. A cost model was developed and consequently various algorithms were created aimed at increasing the probability of detecting a genuine outage, thus lowering the unnecessary costs associated with each outage. Additionally, incoming call frequency was analyzed to detect unusual spikes in customer call patterns. A conservative estimated value added based on part of the deliverables will save Comcast approximately $1.4 million annually.
First Place Winner
Spring 2011