This work proposes a framework for distributed data analytic for precision agriculture based on seamless computing paradigm named SeamFarm. Generally, nodes deployed for precision agriculture consists of different types of devices where these nodes generate an extensive amount of data. Then machine learning can be used to analyze this data for the application of data analytic for precision agriculture. However, most of the IoT devices are resource-constrained devices, which results in poor performance while conducting a machine learning task. Thus, in SeamFarm, we consider to distribute the data as well as the task to all available nodes. The results show that SeamFarm can meet all of the functional and non-functional requirements of distributed data analytic for precision agriculture. It can maintain the resource of device, such as CPU usage during the execution of data analytic application. Moreover, it can obtain faster data analytic results because all of the processes run locally instead of the cloud node.