AHEACS: A Streaming Dataset for Automatic Home Electrical Appliance Control Systems

Introducing a real-time dataset designed for Automatic Home Electrical Appliance Control Systems (AHEACS).Unlike existing datasets with fixed time periods and static nature, AHEACS continuously collects data from an IoT smart home application.This paper presents a method namely Sensor Hub for creating real-time datasets with uses leveraging Message Queuing Telemetry Transport (MQTT) as the communication protocol.AHEACS is constructed and subsequently evaluated by using various machine learning algorithms.The dataset encompasses data from diverse sensors, including smart sockets linked to home appliances, a smart power meter, temperature and humidity sensors, as weel as motion detectors.AHEACS captures appliance ON/OFF status, total household power consumption, and environmental parameters like temperature, humidity, and occupancy.This research explores correlations between variables, particularly the air conditioner's ON/OFF state and other factors.A home automation model for air conditioners is developed based on household power consumption patterns and environmental conditions.The dataset is recorded minutely over a week, accumulating 20,403 data points or equivalent to approximately 2 weeks and 4 hours, stored in .csvformat on a Raspberry Pi 4 edge device.Python in Google Colab is employed for data analysis, including Exploratory Data Analysis (EDA) and six machine learning algorithms: Artifical Neural Network (ANN), Decission Tree (DT), Random Forest (RF), Super Vector Machine (SVM), Linier Regression (LR), and Extreme Gradient Boosting (XGBoost).A case study on the air conditioner's ON/OFF status reveals strong correlations with Total Power (TP) and Presence Bedroom (Pb) variables.ANN emerges as the top-performing algorithm, achieving 86.8% accuracy, 0.097 Mean Square Error (MSE), 0.312 Root Mean Square Error (RMSE), and 0.607 R.