IT Operations Analytics: Using Text Analytics and Statistical Modeling in IT Operations Data
Attendees will be exposed to the emerging area of IT Operations Analytics. Attendees will learn how text mining and statistical modeling techniques can be used to extract insights out of IT Operations Data.
In this talk we will introduce attendees to the up-and-coming area known as IT Operations Analytics (ITOA). A typical large organization with servers, middleware, network switches, and other applications generates hundreds of gigabytes of data each day. This data (known as IT operations data) in most cases is semi-structured in nature and comprises log data, server data, application data, and other machine-generated data. In this presentation we will showcase how some statistical modeling techniques and an IBM product known as Log Analysis were used to derive insights from gigabytes of data. IT operations data also contains ticket data (incident data) with text comments and feedback entered by systems support teams. We employ detailed text pre-processing, clustering, and text analytics algorithms such as LDA and bi-term topic models to indetify topics across support tickets handled by various teams to identify problem hot spots.
I am the Practice Head for Data Science and Engineering in IT Operations Analytics at IBM. I lead a talented team of Data Scientists and Data Engineers. I hold a PhD in Industrial Engineering from the US and have been an active researcher and academician in the US for over a decade. I have given over 26 talks across the world in unsupervised machine learning techniques and analytics with applications in the energy domain.