DEVELOPMENT OF YIELD PREDICTION MODELS IN THE MAIZE CROP USING SPECTRAL DATA FOR PRECISION AGRICULTURE APPLICATIONS

Abstract

Machine learning techniques were applied with statistical tools such as linear, logistic and multinomial regression, to work out predictive algorithms for yield estimation. Spectroradiometer readings were collected throughout the main maiz producing provinces of Ecuador, at two crop development stages. A model using six degree polynomial regression is recommended for acceptable yield prediction. This model could contribute to decide about imports strategies and avoid the overlapping with the national production.

Keywords

• aprendizaje automático; ndvi; sensoramiento proximal; sensoramiento remoto; regresión; machine learning; ndvi; proximal sensing; remote sensing; regression