Interactive Visual Data Mining of a Large Fire Detector Database
SeungJin Lim · 2010
As sensor networks become ubiquitous, the need for data mining of sensor network data is gaining momentum. Sensor network data is typically large, noisy and imbalanced, which makes it challenging to build a robust model from the data. In addition, traditional data mining often requires postmortem processing of the resulting statistically significant patterns to identify interesting patterns by means of visualization. For this reason, interactive visual data mining is employed for mining patterns from the fire detector dataset of the National Fire Incident Reporting System (NFIRS) database in this work. The suitability of interactive visual data mining, in place of its traditional counterpart, is demonstrated.