Predicting the Number of Failures by Using Field Data

In this study, our aim is to predict the future number of failures by analyzing the field failure data of a collection of complex repairable systems and we propose a method to accomplish this goal. In the proposed method, first we find the Mean Cumulative Function (MCF) of failures on monthly basis and then we fit a curve to this function. We select the best fit based on two criteria as Leave-one-out Cross Validation (LOOCV) and Root Mean Squared Error (RMSE) of the last years' values. We apply this method to two different product types and detect the best fits. Finally we interpret our results and share our experiences that we gain throughout the study.