Lead Data Scientist
Exploiting small and big data for machine learning based incident prevention
|Device failures are unavoidable where large numbers of different equipments are in operation to serve customer needs. This is true to British Telecommunication, too, that provides landline internet services to thousands of enterprise customers all around the globe. It is in everyone’s interest to react and repair any problems as fast as possible. Or even better, anticipate the future problems and prepare for them.
The presentation is focusing on two approaches experimented at BT Business that give analytical support to the incident management teams.
The second approach is processing large amount of sensor data in a near-real-time fashion and trying to spot indicators of an upcoming failure on the device level. As at the time of writing this synopsys the project is still in an experimenting phase, more focus can be promised to put on the immediate challenges with the sheer management of the immense volumes of the source data and how the toolset of Google Cloud can support it. Those interested in the analytical aspects can be given some guidance as well on how popular textbook methods fared with the problem when used just naively.
Tamás Molnár has been working as data scientist for one and half decade. He has delivered insights and predictive models in the telecommunication, financial and healthcare sector. The supported domains have been very diverse from sales-marketing, finance, risk management and SCM to HR and process analysis.