Fruit yield prediction of pepper using artificial neural network

• Published in: Scientia horticulturae 2019-05, Vol.250, p.249-253
• Main Authors: Gholipoor, Manoochehr, Nadali, Fathollah
• Format: Article
• Language: English
• Subjects: Canopy; Flowering; Fruit number; Modelling
• ISSN: 0304-4238; 1879-1018
• DOI: 10.1016/j.scienta.2019.02.040

•Pepper fruit yield response to participant traits was quantified precisely (R2=0.97) by artificial neural network (ANN).•ANN Architecture of 8:10:1 tended to be best, when log-sigmoid transfer function was employed.•Fruit number per plant had highest participation to the determination of fruit yield. The prediction of pepper fruit yield is one of the most important breeding objectives in agricultural research. So, in this study, the artificial neural network (ANN) was employed to quantify the pepper fruit yield response to traits days from sowing to emergence, days from sowing to flowering initiation, days from sowing to 50% flowering, plant height, canopy width, number of fruits per plant, fruit water content, and reproductive stage duration. In an outdoor experiment in Shahrood, Semnan province, Iran, the seeds of 692 indigenous genotypes of hot pepper (Capsicum annuum L.) were sown to get the data for establishing and testing ANN. The results indicated that ANN with architecture of 8:10:1 achieved the highest accuracy (R2 = 0.97). Finally, sensitivity analysis was conducted to determine the relative contribution of plant characters to the determination of fruit yield. In terms of relative contribution, they ranked from higher to lower as the fruit number per plant, days from sowing to 50% flowering, days from sowing to flowering initiation, reproductive stage duration, fruit water content, canopy width, days from sowing to emergence, and plant height.