LoRa-Based Waste Bin Monitoring for Making Decision Waste Disposal Using C 4.5 Method

Historical data on waste management activities should be used to assist decision-making. This study aims to make past or multi-sensor historical data from smart waste bins a reference for making decisions about collecting waste from smart waste bins. This research contributes to producing a method to make historical data on waste management in rural areas become the rules in making decisions about waste disposal as appropriate. Quantitative waste volume data were obtained from a LoRa-based smart waste bin specifically using the E32-TTL-1W series LoRa network media, each unit with three parameters, namely the volume of metal, organic and inorganic waste. The main sensor devices used in this research are Capacitive Proximity sensor LJC18A3-H-Z/BY, Inductive Proximity sensor LJC12A3-4-Z/BY, Ultrasonic sensor HC-SR04. Waste volume data from monitoring activities of smart waste bins and disposal decision data by officials are converted into historical data. Historical data in the quantitative form is converted into categorical data in the form of full, filled, and empty, then extracted into rules using the C 4.5 method. A waste monitoring system is obtained that can produce waste disposal decisions from smart waste bins. The test results obtained an accuracy value of 84.85 percent for method C 4.5, Randomforest at 84.85 percent, and Randomtree at 76.52 percent. The system can be used to provide decision recommendations from multiple input sensors for other systems.