Coffee is one of the main plant-based goods consumed today, as well as the second most traded commodity in the modern economy after oil. Its production takes place mostly in developing countries such as Colombia, where a malfunction during the process could affect thousands of coffee farmers. Coffee rust does just that, reducing the quality and quantity of grains devastating entire productions yearly. One of the most effective ways to prevent this disease from spreading is early detection, which makes effective the extraction and replacement of withered crops. Using wireless sensor networks, it is possible to obtain data from each plant’s conditions; which allows us to determine whether a specific plant is in the early stages of infection or perfectly healthy. Serving this purpose, a data mining and machine learning concept called decision trees was implemented. With the help of one of these alone, we were able to correctly classify 79% of the plants that were studied. In search of a better result, a new method was used; one called random forest where multiple trees are built and the most common result is taken as the final one.