This talk will cover an example of predicting remaining useful life of turbofan jet engines based on measurements from sensors and operating settings. This type of prediction is a common use case for machine prognostics and health management. In this talk we will demonstrate the use of the open source machine learning platform, H2O, for data preparation and modeling, some tips on cross validation strategies for time-series data, and the use of Kalman filters for post-processing of predictions.
Hank Roark is a Data Scientist and Hacker at H2O. Hank has a broad background in physics, software engineering, research, and data science. Hank's particular enjoys modeling data from sensors and the Internet of Things. He dreams in dictionary comprehensions and likes pop-up TMBG concerts.