Multi-Class Classification is a central task in data mining there are a variety of classifiers that compute scores of the samples to belong to the classes. The aim is to deliver a product that will be used to analyse the classification algorithms, find sources of classification errors and reducing errors using available data features. This involves evaluating performance of classifiers using Interactive Visualizations and providing the users with useful and reproducible information about classification Data. Back-end plugins for WEKA, RapidMiner, KNIME and R to acquire and export classification data to front-end that produce visualizations using web technologies (D3.js) in form of stacked bar charts and confusion matrix, which will easy, intuitive and high quality implementation to interactively select certain samples in the visualization for further investigation and detailed feature analysis run in default browser to provide platform independent functionality. A guide and documentation for all the code for easy installation, usage, interpretation of results and further extension of project related to this tool in the form of a web-page will be a part of this project.




  • Emma Beauxis-Aussalet