2020 International Conference on Electrical, Communication, and Computer Engineering (ICECCE) · 2020 · 11 citations · 23 references
ComputingComputer ScienceArtificial IntelligenceForecastingTransportationIntelligent Transportation SystemsNeural NetworksReal-time ForecastingSmart CitiesTime Series ForecastingTime Series PredictionError PredictionMachine Learning-based ForecastingMean Absolute ErrorNeural ForecastingReal-time Traffic ForecastingTraffic Flow ForecastingTraffic PredictionTravel Time Prediction
The concept of Smart Cities is a trend in large cities. Intelligent Transportation Systems) plays an essential role in providing accurate information on bus travel times. It improves the planning of passengers and the agency responsible for public transport. The purpose of this paper is to create a new methodology to predict the travel times of buses based on open data collected in real-time. We constructed a dataset from Sao Paulo City bus fleet location data, real-time traffic data, and traffic forecast from Google Maps. In the following, we trained an Artificial Neural Network (ANN). In the ANN training process, we alternated the dataset and its hyperparameters to find out the combination that provides the lower prediction error. The mean absolute percentage error found was 8.97%, using all data sets except weather data. We showed that our method could provide an accurate bus travel time prediction from web data collected in real-time.
23