Sistema de recomendação de plantio utilizando aprendizado de máquina

Abstract

Through computer systems for managing farms, drones, autonomous vehicles, humidity and temperature sensors, there has been a huge growth in the amount of data coming from agronomic activities. This data then began to be used in Machine Learning algorithms in order to seek greater planting yield, harvest prediction, recognition of plant diseases, planting recommendation among other applications. This work aims at the development of creating and sharing a planting recommendation system using Machine Learning algorithms, in order to facilitate the choice of the best crop according to the database used, which has the characteristics of the place, such as: humidity, temperature, pH and nutrients present in the soil. For the creation and sharing of this planting recommendation system, algorithms aimed at data classification were used, such as Decision Tree, Random Forest, Naive Bayes, among others. The models were created using the Python programming language and shared through the Streamlit library. Thus, after creating and comparing the models, it was incorporated into the web application, created through Streamlit, the Random Forest algorithm. In the end, it was possible to create an application with excellent recommendation performance, intuitive and easy to use for the user.