Visual Data Analysis

Abstract

Visual techniques were used to support complex customer data analytics and illustrate concepts and data mining algorithms behind the built Recommender System. The implemented web-based visual system supports a feature analysis of 38 client companies for each year between 2011–2016 in the area of customer service (divided into shop service/field service) and parts. It serves as a visualization for the feature selection method showing the relative importance of each feature in terms of its relevance to the decision attribute—Promoter Status, and is based on an algorithm that finds minimal decision reducts. It allows the user to interactively assess the changes and implications onto predictive characteristics of the knowledge-based model. It is supported by visual additions in form of charts showing accuracy, coverage and confusion matrix of the model built on the corresponding, user-chosen dataset.