Reliable Yield Prediction with Regression Neural Networks

This paper performs an extension to conventional regression neural networks (NNs) for replacing the point predictions they produce with prediction intervals that satisfy a required level of confidence. Our approach follows a novel machine learning framework, called Conformal Prediction (CP), for assigning reliable confidence measures to predictions without assuming anything more than that the data are independent and identically distributed (i.i.d.). We evaluate the proposed method on predicting the weekly yield of tomato in a greenhouse, two datasets were used, which consist of environmental variables inside the greenhouse, namely temperature, CO2, vapour pressure deficit (VPD) and radiation, as well as past yield. Greenhouse environment data and crop records were collected from a large scale commercial operation, Wight Salads Group (WSG) in the Isle of Wight, United Kingdom; we use a dataset of more than 60000 record measurements collected over a period of 11 years. Our experimental results show that the prediction intervals produced by our method are both well calibrated and good enough to be useful in practice.