Wireless sensor networks are of energy-constrained nature, which calls for energy efficient protocols as a primary design goal. Thus, minimizing energy consumption is a main challenge.We are concerned in howcollected data by sensors, can be processed to increase the relevance of certain mass of data and reduce the overall data traffic. Since sensor nodes are often densely deployed, the data collected by nearby nodes are either redundant or correlated. One of the great challenges for the aforementioned problem is to exploit temporal and spatial correlation among the source nodes. Our work is composed of two main tasks: 1- A predictive modeling task that aims to capture the temporal correlation among collected data. 2- A data similarity detection task that measures the data similarity based on the spatial correlation.