Improvised KDE-Track Based Efficient Dynamic Evaluation System for Data Stream Cluster

Now days the availability of the time and space of the data stream are increased due to the development to the sensors and the global positioning system development the system. Its very difficult task to the develop the model can mine the data stream with the such extensibility Its also difficult the analyze of the incoming data because the data streams Rate is very fast. Current state of art systems are either expensive or inaccurate in capturing changes in incoming data streams Also there can be an occurrence of dynamic changes in data distribution and is very important to be identified. To overcome these drawbacks a new system is proposed named KDE-track for which can keep a track of such spectrometric data streams. The system consists of an adaptive re-sampling which can solve the issue of inaccuracy in capturing changes in incoming data streams and thus, can estimate density dynamically. Also a new method is proposed for accurate and efficient selection of bandwidth for kernel density estimator. The experiments show KDE-track is leaving task for mining such data streams due to its state-of art systems in terms of performance.