red_leaf07
Registered
I am working on a dataset that has records of machines tested until failure. As usual in testing environments, we don't necessarily have complete data with machines running until failure since the test duration is fixed and limited.
So my idea is to apply survival analysis to not ignore the censored data but I am facing some road blocks in try to understand the how to predict time to failure and the how to intepret the results. I have compiled the data in such a way that I have the machine IDs, the duration until failure or until test ended if censored and status whether dead or censored. There are some covariates as well and when I tried to apply Cox regression, I realised that the prediction value I was getting were risk rates rather that time to event. Since I am coming from mainly data science approach, I am not very familiar with using survival analysis but I am still interest to leverage its usefulness especially in dealing with censored observations.
I would appreciate some insights and advice on how to proceed and how to properly apply the models and also on the required hypothesis testing. I am using mainly R and python to work on this. Thanks.
So my idea is to apply survival analysis to not ignore the censored data but I am facing some road blocks in try to understand the how to predict time to failure and the how to intepret the results. I have compiled the data in such a way that I have the machine IDs, the duration until failure or until test ended if censored and status whether dead or censored. There are some covariates as well and when I tried to apply Cox regression, I realised that the prediction value I was getting were risk rates rather that time to event. Since I am coming from mainly data science approach, I am not very familiar with using survival analysis but I am still interest to leverage its usefulness especially in dealing with censored observations.
I would appreciate some insights and advice on how to proceed and how to properly apply the models and also on the required hypothesis testing. I am using mainly R and python to work on this. Thanks.