Multivariate Data Fusion for Traffic Prediction in Smart Villages

In the context of the Internet of Things (IoT), this paper identifies the data engineering and analytical challenges of traffic prediction in smart villages arising from the diversity of devices and heterogeneous data sources. It considers the integration of diverse temporal data streams, specifically Automatic Number Plate Recognition (ANPR) cameras for vehicles, people-counting sensors in urban areas, and social data from web search trends. As a line of work, we propose a scalable methodology that combines robust data pipelines with a multivariate time series approach. This approach incorporates causal discovery and techniques based on mutual information and entropy for feature selection. The objective is to establish a reproducible framework that efficiently fuses multi-structured information from disparate devices, thereby mitigating noise and optimizing traffic prediction accuracy in rural environments.