Due to enormous growth in mobile device technology, user's preferences have been shifted from traditional mobile phones and laptops to the other handheld devices such as smartphones. Signifícant efforts have been made to make smartphones rich in terms of processing capabilities and reduction on energy consumption. Despite the improvements in provision of computational, memory and energy resources with smart phones, Smart phones are still characterized as resource constrained devices. It is believed that increasing resource capabilities in smart phones cannot handle exponential increase in smart phone applications and the resultant network traffic. Cloud computing has emerged as viable solution to address the user's increasing resource requirements. To achieve computational efficiency in terms of speed, recent researches have recommended that programming codes that require intensive computational resourcescan be offloaded to the cloud servers. However, the accuracy of decision to offload code to cloud server can largely impact the performance of the overall system. In this paper, we propose an accurate decision making system for adaptive and dynamic nature of mobile systems by using Support Vector machine learning technique for making offloading decision locally or remotely. Proposed system is evaluated with Android-based prototype component for experiments considering different internal and external conditions (network characteristics). Our proposed system achieves approximately 92% accuracy, leading to accurate decision, thus improving performance and reducing energy consumption. © 2016 IEEE.