Application Analysis of Supervised and Unsupervised Methods for Clustering and Classification of a Wheat Seeds Dataset

Abstract

This project aims to apply easy-to-implement supervised and unsupervised learning methods to do a cluster and classification analysis of the information from a wheat seeds dataset. In addition, it intends to thoroughly evaluate, employing post-processing techniques, the efficiency of the models produced from them, proving that it is not necessary to use complex procedures in this database, since the obtained results for clustering and classification were highly similar to the dataset original labels. The decision tree was chosen as the classification algorithm. Furthermore, k-means and Kohonen neural network were selected as the clustering methods. Additionally, pre-processing and exploratory data analysis techniques were explained in detail and employed in order to maximize the final quality of the model developed.