Effect of Time Intervals on K-nearest Neighbors Model for Short-term Traffic Flow Prediction

摘要: The accuracy and reliability in predicting short-term traffic flow is important. K-nearest neighbors (K-NN) approach has been widely used as a nonparametric model for prediction. However, the of K-NN results unknown uncertainty point prediction needs to be quantified. To this end, we extended by constructing interval associated with Recognizing stochastic nature traffic, time measure rate remarkably influential. In paper, extensive tests have also conducted after aggregating real data into intervals, ranging from 3 minutes 30 minutes. show that performance can improved when increases. More importantly, shorter than 10 minutes, generate higher selected benchmark model. This finding suggests may more appropriate at interval.