0
$\begingroup$

We have sensor data (temperature, pressure) for about 80 sensors and am wanting to be able to recognise when a component 'failure' is occurring to shut down in advance. Predicting the next failure would be a secondary goal.

What would be a basic analytic approach here. I have read about PCA to reduce the dimensionality of the data and that seems to be something sensible to do as there are too many variables to start with. After PCA what am I then doing? - some form of classification/regression of failure status (y) on the main components (x)?

But these are also time-series data, right? Is it important then to model time as well? I'm completely new to time-series analysis and don't really understand what it's purpose is.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.