I have hardware data with a few dozens of features, covering a few machines over a few years, and a few hundreds of failures. My task is to predict future failures. Does anyone has experience with such failure prediction problems? If so, can you direct me to some good resources for this problem?