Authors:   M. Olaniyan, A. B. Owolabi

Abstract

This study was carried out to predict the extent of postharvest loss in three agrifood commodities namely rice, maize and yam along the food value chain in Delta State, Nigeria. The study considered famers, transporters, processors, marketers and consumers as the five principal actors in the value chain with farmers being the harvester. Sufficient relevant information was obtained from each of the actors with the aid of organized interviews and well- structured questionnaires. The questionnaires contain information relating to postharvest loss in each of the three commodities at every stage in the value chain - from harvest to consumption. 450 questionnaires were administered on each commodity, with 150 being handled by each actor in each commodity in each of the three senatorial districts in the state making a total of 2250 questionnaires that were administered altogether. Five types of Artificial Neural Network (ANN) topology were used for each commodity making a total of fifteen models that were used for three-layer feed-forward model (TL-FFM) with back-propagation multi-layer perception (BP- MLP) type of ANN. Data analysis was carried out by ANN-ALYUDA forecaster software under the TL- FFM with BP-MLP. Result obtained showed that transporters, processors and marketers contributed more to postharvest loss in rice, maize and yam compared with farmers and consumers. It can be inferred from this study that ANN using TL-FFM with the supervised training type BP-MLP is one of the best tools that can be used to predict postharvest loss in any agricultural commodity along the food value chain. This is due to its understanding in learning the pattern the input data followed and hence predict accurately the target output with little deviation and minimum error. Comparison between predicted values and the target output values in each of the fifteen models showed how good the ANN had been trained to predict losses that occurred along the value chain based on the five actors that contributed to postharvest loss in each commodity.