Comparison of predictive accuracy and descriptive power of Machine Learning algorithms through a case study involving maintenance of jet engines
Jet engines are complex machines, subject to demanding operating conditions, whose proper maintenance is critical to ensure high safety, maximum availability and minimum lifecycle costs. Therefore, dozens of sensors measure in real-time hundreds of engine parameters (temperature, pressure, vibration, shaft rotation speeds…) to monitor and predict the deterioration of jet engines.
We present some of the results obtained during a PhD thesis held at the Instituto Superior Tecnico, in partnership with Rolls-Royce and MIT. After introducing the context and case study, we briefly present the framework designed during the thesis to analyse the complex jet engine data. The second part of the presentation will compare the predictive accuracies of the Machine Learning techniques selected to predict engine deterioration (logistic regression, Naive Bayes, random forests…). The third part will cover the descriptive ability of these machine Learning techniques, notably via the notion of marginal dependence of the predictors on the response variable.