Predição de ranqueamento de lotes de sementes de milho por inteligência artificial

Abstract

In the seed sector there are several challenges in making quick and accurate decisions when working with huge quantities of seed lots, with manual handling becoming almost impossible. The reinvention of agriculture comes with artificial intelligence (AI), mainly in the quality sector. The objective of this work is to use machine learning to classify corn seed lots. Data from a corn seed company were used, with the attributes being eight harvests. The classifiers used were J48, RandomForest, CVR, lBk, MLP and NäiveBayes. Cross validation was used, in which the data set, training and testing, was divided into 10 subsets. The steps described were performed using the Weka software. The most accurate data mining technique was prediction and classification (100%) and not the association technique with only 70.35% accuracy. It is possible to classify lots of corn seeds with great accuracy and precision through artificial intelligence and its machine learning technique. Artificial intelligence will allow you to automate processes in the company and increase the speed of completing tasks.