Self-organizing Data Processing for Time Series Using SPARK

Abstract

Time series data can be found at many places in our day-to-day life. As the world is moving toward Internet of things (IoT) and sensors, data is obtained in enormous amounts. These data from multiple sources need to be combined as a single unit in order to make inferences or predictions. As the velocity of data generation increases, big data perspective needs to be applied to time series. Big data time series representation should be flexible enough to accommodate different time series. With multiple time series collected from different sources and at different intervals, it poses a challenge to combine these time series into a single one to perform analysis. In this paper, we propose a big data approach to represent time series that provides solution to these challenges and helps in analysis of the data. The proposed approach provides an efficient way to combine multiple time series and perform analysis on it.