I am trying to predict the time to certain type of failure given the following data on Certain Factory Equipment. The data I have are readings collected every day for sensor installed on those equipment. On Same day, an equipment can have different Repairs performed,and on some days no repairs. When a repair is performed it means some part of that equipment failed.

Our goal is to find the time to failure of that part so we can perform preventative maintenance. I converted the data so I have only one type of failure in my data , Part1 failed or not failed is my binary output. I intend to do seperate analysis for different part failures.

I was thinking of using the randomForestSRC package to fit a model. I have not done any survival Timeseries Modeling. My challenge is how to define the Input Time series in the surv(time,output) What other things should be kept in mind when doing survival analysis on such data?

Can I just sort the data by time and use that as input or I need to define my Starttime (time when the equipment was brand new and stop time as time when the equipment had certain part failure?)

Should the analysis be done separately on each Equipment ID ? Any ideas appreciated.

Sample data: enter image description here



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.