Home Automation Based on User Behavior for Air Conditioning Electronics Devices with Classification Using Random Forest Algorithm

A smart plug is a device designed to control electronic appliances in home automation remotely. One of the most essential appliances in a household is an air conditioner (AC). AC is necessary during very hot weather and consumes a significant amount of power. It can be challenging for families with limited power capacity. The objective of this study is to control AC appliances automatically to limit electricity consumption. Automation is implemented on the smart plug by controlling the AC's on/off status based on temperature and humidity sensors, motion sensors, total household power consumption, time, and the status of the water pump. Data generated by the smart plug is combined with other sensor data. Then, the combined data is processed using a big data pipeline with an Edge Server, namely Node-RED. Node-RED is used to manage data in near real-time. Data is ingested into Node-RED, then stored in MongoDB Cloud Atlas, and pre-processed into a dataset for classification for status AC. Sensor data and electronic device statuses are displayed on the home automation dashboard. The Dashboard is developed using the Laravel framework and Rubix ML. Rubix ML is used for implementing the Random Forest classification model of AC status, which achieved an accuracy of 78%. The dashboard provides near real-time information about energy usage, AC status, and estimated electricity costs. The predictions displayed on the dashboard were successfully tested and validated through direct testing of the smart plug AC with users, yielding satisfactory results.