This study develops a predictive model using machine learning techniques to estimate banana bunch weight in Mata de Cacao precinct, Los Ríos province, Ecuador. Exploratory and descriptive analysis methods were used to identify relevant variables, such as top calibration, number of hands, week and minimum temperature. Since, there is no previous research findings concerning bunch weight estimation through inferential statistics techniques in this geographic region, five different regression algorithms were evaluated: Boosted Gradient (GB), Ridge (Ridge), Nearest Neighbors (KNN), Least Absolute Shrinkage and Selection Operator (LASSO) and Light Gradient Boosting Machine (LightGBM). Among these, LightGBM proved to be the most effective, with a coefficient of determination (R2) of 0.85, a mean square error (MSE) of 50.43 and a mean absolute error (MAE) of 4.21. In comparison, Ridge and LASSO showed lower performances, with R2 of 0.69 and 0.66. The integration of the model into the farming web application helps to reduce management costs and labor dependency by means of traditional methods.