Logged data from business processes contain a treasure trove of information that usually remains untapped. Now that businesses have many more instrumented processes with IoT devices, the data collected is bound to increase. In many cases, businesses collect the data to answer: 'what happened?' However, this information could also be used to anticipate what is likely to happen. In particular, we could use the data collected to determine how long a given process is expected to take to complete.
In this presentation, I will present a case study on building a continuous learning algorithm to predict how long an Ansible playbook will take to execute. Ansible is a tool that IT organizations use to automate tasks. A developer could go for a cup of coffee or out to lunch only if they know how long the build would take. The historical data stored in Logstash (from the Elastic Stack) will be used to feed a k-Nearest Neighbor algorithm for predictions.
Zeydy Ortiz, Ph. D. is the co-founder and CEO of DataCrunch Lab. She helps teams and organizations translate data into actionable insights for operational efficiency. Her interests include incorporating artificial intelligence and machine learning into Internet of Things applications. She is also the organizer of NC Data4Good bringing together social good organizations and data scientists to make an impact in our communities. Dr. Ortiz has a bachelor's degree in Computer Engineering from the University of Puerto Rico, a Master's in Computer Science from Texas A&M University and a doctorate from NC State University. Before focusing on data science, she was a performance engineer at IBM developing predictive models to forecast the performance of servers. She lives in Cary with her two kids, two dogs and her husband.----