Time-fading based high utility pattern mining from uncertain data streams

Abstract

Recently, high utility pattern mining from data streams has become a great challenge to the data mining researchers due to rapid changes in technology. Data streams are continuous flow of data with rapid rate and huge volumes. There are mainly three widow models namely: Landmark window, sliding window and time-fading window used over the data streams in different applications. In many applications knowledge discovery from the data which is available in current window is required to respond quickly. Next the Landmark window keeps the information from the specific time point to the present time. Where as in time-fading model information is also captured from the landmark time to current time but it assigns the different weights to the different batches or transactions. Time-fading model is mainly suitable for mining the uncertain data which is generated by many sources like sensor data streams and so on. In this paper, we have proposed an approach using time-fading window model to mine high utility patterns from uncertain data streams. © Springer International Publishing Switzerland 2014.