Nowadays Big Data has a key role in E-Agriculture. Previous technologies have some limitations so Big Data is very useful for E-Agriculture because agriculture has very large structure and unstructured data. Moreover, big data analytics can be used to increase and improve the productivity of agricultural. The main aim of this paper is to propose an open source, economical, ideal and flexible big data analytics architecture for E-Agriculture. In the implementation, an analytic framework for big data application development is built and implemented. Also, a prototype application prediction base for agriculture in spark framework. Based on the agriculture prediction model various recommendations can be provided to agro users.