A Simple Dense Pixel Visualization for Mobile Sensor Data Mining

• Tipo: Artigo em Livro de Atas de Conferência Internacional
• Data: 2010

Autores

Pedro Pereira Rodrigues (Autor) FMUP; Joao Gama (Autor) FEP

Ata de Conferência Internacional

Título: Knowledge Discovery from Sensor Data

Páginas: 175-189

Indexação

ISI Web of Knowledge - 2 Citações

Scopus - 4 Citações

Classificação Científica

FOS: Ciências da engenharia e tecnologias > Engenharia electrotécnica, electrónica e informática

Outras Informações

ID Authenticus: P-003-BTK

DOI: 10.1007/978-3-642-12519-5_10

Abstract (EN): Sensor data is usually represented by streaming time series. Current state-of-the-art systems for visualization include line plots and three-dimensional representations, which most of the time require screen resolutions that are not available in small transient mobile devices. Moreover, when data presents cyclic behaviors, such as in the electricity domain, predictive models may tend to give higher errors in certain recurrent points of time, but the human-eye is not trained to notice this cycles in a long stream. In these contexts, information is usually hard to extract from visualization. New visualization techniques may help to detect recurrent faulty predictions. En this paper we inspect visualization techniques in the scope of a real-world sensor network, quickly dwelling into future trends in visualization in transient mobile devices. We propose a simple dense pixel display visualization system, exploiting the benefits that it may represent on detecting and correcting recurrent faulty predictions. A case study is also presented, where a simple corrective strategy is studied in the context of global electrical load demand, exemplifying the utility of the new visualization method when compared with automatic detection of recurrent errors.

Idioma: Inglês

Tipo (Avaliação Docente): Científica

Nº de páginas: 15