Authors:
David L. García
;
Àngela Nebot
and
Alfredo Vellido
Affiliation:
Universitat Politècnica de Catalunya - Barcelona TECH, Spain
Keyword(s):
Visualization, Cartogram, FlowMaps, Generative Topographic Mapping, Churn, Telecommunications market.
Related
Ontology
Subjects/Areas/Topics:
Abstract Data Visualization
;
Computer Vision, Visualization and Computer Graphics
;
Databases and Visualization, Visual Data Mining
;
General Data Visualization
;
High-Dimensional Data and Dimensionality Reduction
;
Visual Data Analysis and Knowledge Discovery
;
Visualization Algorithms and Technologies
;
Visualization Applications
Abstract:
Telecommunication companies compete in increasingly aggressive markets. Avoiding customer defection, or churn, should be at the core of successful management in such context. These companies store and manage abundant customer usage data. Their analysis using advanced techniques can be a source of valuable insight into customers’ behavior over time. Exploratory data visualization can help in this task. Many important contributions to multivariate data visualization using nonlinear techniques have recently been made. In this paper, we analyze a database of customer landline telephone usage in Brazil. These data are first visualized using a nonlinear manifold learning model, Generative Topographic Mapping (GTM). This visualization is enhanced using a cartogram technique, inspired in geographical representation methods, that reintroduces the local nonlinear distortion into the representation space. Yet another geographical information visualization technique, namely the Flow Maps, is the
n used to visualize customer migrations over time periods in the GTM data representation space. The experimental results shown in this paper provide evidence to support that the use of these methods can assist experts in the process of useful knowledge extraction, with an impact on customer retention management strategies.
(More)