A classifier based on KNN and KMeans methods.
Authors: Pau Blasco Roca, Biel González Garriga, Alberto Real Quereda, Cristina Soler Arenys
Date of release: 23/5/2022
Date of latest update: 22/6/2022
Classification methods allow us to classify elements from a dataset into different classes, based on their characteristics. This project focuses on using various methods of classification in a database of clothing samples to categorize them by color and shape (or type). This has led to a number of potential improvements and subsequent analysis of these for reliability and speed. Subsequently, the performance change was studied for each one of the different improvements to visualize and understand their impact. In this way, it has been possible to assess whether the changes have originated significant improvements, or not.
Note: Parts of this project, aswell of the main analysis ("Informe"), are written in Catalan.
Patches:
1.0: Original project
1.1: Added a verbose option to KNN's predict show extra info about the voting system