K-Nearest Neighbors for Smart Solution Transportation: Prediction Distance Travel and Optimization of Fuel Usage and Charging Recommendations for ICE Vehicles Based in Surabaya

Surabaya ranks 9th in Southeast Asia and 44th globally in the TomTom Traffic Index, with an average travel time of ±22 minutes for a 10 km distance, longer than Jakarta’s ±20 minutes. Given these traffic conditions, this study examines the application of the K-Nearest Neighbors (KNN) algorithm to predict vehicle travel distance based on remaining fuel consumption and provides recommendations for the nearest Gas Station (SPBU) based on the predicted distance. The study seeks to provide accurate distance predictions and recommend the nearest Gas Station (SPBU) for users based on fuel consumption and the predicted route, helping to navigate Surabaya’s congested traffic efficiently. The data used includes various levels of fuel consumption: 0.02, 0.06, 0.10, 0.14, 0.16, 0.20, and 0.24 liters for engines of 110, 125, and 150 cc. The model evaluation results, using three metrics: MAE, MAPE, and RMSE show that KNN performs excellently at low fuel consumption levels. At a consumption rate of 0.02 liters, the model produces a low MAE of 0.347, MAPE of 31.21%, and RMSE of 0.40, indicating minimal prediction error. The model's performance remains consistent at a consumption of 0.06 liters with MAE of 0.330, MAPE of 9.90%, and RMSE of 0.41, demonstrating a high level of accuracy. Technically, the implementation of this model can help reduce traffic congestion by directing vehicles to the nearest gas stations, thereby minimizing sudden stops on the road, improving traffic flow, and reduce wasted time spent searching for distant gas stations.