The emergence of Internet of Things (IoT) has delivered a new perspective about how technology can accomplish human needs, IoT primarily used to deliver rapid stream of data from sensors to variety options of storage system and used it to accomplish objective of the development. As a result of that, IoT may occurs an unavoidable challenges of data noise. In this paper we will present our result of data noise removal approach based on data stream of IoT scale machine that we built. Outliers and window functions will be used to pre-processing the raw data of IoT, by applying this approach, we aim to deliver a better data in the pre-processing phase as an effort to eradicate the data noise problem in IoT sensors. IoT device that we built was using NodeMCU microcontroller, HX711 amplifier and 200 kg loadcell. We also comparing Exponential Moving Average (EMA) to other Window Function to deliver an insight for reader of EMA performance across other window function.