Pedestrian flow prediction in commercial avenue

Mobility plans are one of the most important management tools for city development and an important factor for society and economic growth, where pedestrians are the end goal of any mobility plan. Human behavior is generally unpredictable, and many attempts have been interested at pedestrians' mobility in urban environments, both microscopic and macroscopic (flow, density, and speed) levels. The objective of pedestrian traffic flow prediction is to predict the number of pedestrians at the next moment. Assisting operators and city managers in making decisions in urban environments such as emergency support systems, and quality-of-service evaluation. This study aims to model and predict bi-directional pedestrian flow in a commercial avenue, based on two essential stages, data collection through video recording over two months (pedestrian flow) and data analysis using machine learning algorithms that provide a lower error and a higher accuracy rate. Two metrics were selected as basic measures to evaluate the model performances, root mean square error (RMSE) and coefficient of determination R2. Artificial neural network (ANN) gives a little better performance and fitness.