<span>The main interest in many research problems in polymer bio composites and machine learning (ML) is the development of predictive models to one or several variables of interest by the use of suitable independent inputs or variables. Nevertheless, these fields have generally adopted several approaches, while bio composite behavior modeling is usually based on phenomenological theories and physical models. These latter are more robust and precise, but they are generally under the restricted predictive ability due to the particular set of conditions. On the other hand, Machine learning models can be highly efficient in the modeling phase by allowing the management of high and massive dimensional sets of data to predict the best behavior of bio composites. In this situation, biomaterial scientists would like to benefit from the comprehension and implementation of the powerful ML models to characterize or predict the bio composites. In this study, we implement a smart methodology employing supervised neural network models to predict the bio composites properties presenting more significant environmental and economic advantages than composites reinforced by synthetic fibers.</span>