Case Study
Railcar Wheel Failure Forecasting For A Largest Railcar Provider In North America
Business Problem The Client wanted to forecast the number of wheel failures in a given year by wheel size and fleet type to be used for their budget and inventory planning. Methods based on simple regression led to large forecast errors due to the presence of censored da- ta. The current practice of dealing with censored da- ta was ad-hoc and led to forecast errors of 7% - 10%, an unacceptable level in this industry. Challenge We had to w