SMU Data Science Review SMU Data Science Review
Using the previous years of parts sales data for a supplier to the oil and gas industry in North America, a novel method to predict demand with a minimal error rate is found.
Abstract
. In this paper, we present a model and methodology for predicting the following quarter’s product sales volume. Forecasting product demand for a single supplier is complicated by seasonal demand variation, business cycle impacts, and customer churn. Based upon a Dense Neural Network (DNN) machine learning model, we created a prediction model that considers cyclical demand variations and customer churn. Using the previous five years of parts sales data for a supplier to the oil and gas industry in North America, we found a novel method to predict demand with a minimal error rate [MAE of 0.65]. The Dense Neural Network model performs the best among the other machine learning models we tried in prediction, and additionally, all machine learning models perform better than a non-machine learning solution.