The phenomenal rise of Internet of Things(IoT) has driven the ascend of fog computing as a distributed model to reduce delays in networks. Generally, fog devices are resource-restricted while IoT applications are becoming computationally demanding which require certain QoS to be accomplished within hard time limits. Therefore, it is preferable for IoT jobs to finish their processing within deadline limits by producing approximate results than producing accurate result late. The placement of IoT jobs on fog and cloud resources for execution is a widely-recognized NP-hard problem. We study the placement of real-time IoT workflows in a fog and cloud infrastructure by applying approximate computations and TOPSIS. Our methodology aims to place complete as well as partial tasks in the available idle schedule holes in the schedules of fog as well as cloud resources. The proposed technique is verified through simulation experiments and is contrasted with state-of-the-art techniques on different performance measures. The experimental findings establish that the proposed technique is able to render better performance compared to its alternatives in terms of performance metrics like SLA violation ratio, response time and energy consumption at an insignificant loss of 0.15% of result precision for all experimental scenarios that are taken into consideration.