Real-Time Probabilistic Approach for Traffic Prediction on IoT Data Streams

Abstract

IoT data analytics refers to the analysis of voluminous data captured by connected devices. These devices interact with the environment and capture the details which are streamed to a central repository where the processing of this data is done. This collected data may be heterogeneous in nature, as research has identified weather, social, and pollution data as key players in traffic prediction in smart cities, making the analytics challenging. In this work, we propose Unir, an event driven framework for analyzing heterogeneous IoT data streams. In the first step, we ingest the data from Twitter, weather and traffic APIs and store it in a persistent data store. Later, this data is preprocessed and analyzed by deep learning models to forecast future data points. In the second step, a supervised Hidden Markov Model consumes the sequence of predicted data points from the first layer. The HMM is trained using the ground truth labels obtained from TomTom API to find the likelihood values of a congestion event. The likelihood for congestion and non-congestion sequences is learned by a Logistic regression which assigns a confidence to the occurrence of an event. The proposed approach displays a 77% overall accuracy in comparison to the baseline approach on experiments.