Bike-sharing Usage Prediction Based on PCA-GABP Network Model

Abstract

The emergence of bike sharing has facilitated our lives, but at the same time, there are problems such as spatiotemporal imbalance. Considering the influence of various natural and human factors on the usage of shared bicycles, this paper selects the data on shared bicycles: in Seoul for a period of time. It combines the principal component analysis method to reduce the comprehensive weights of evaluation indexes. Then, it uses the reduced feature information as the input layer of the GA-BP neural network, which constructs a PCA-GA-BP network model with an accuracy rate of 95%. It conducts a comparison with various other models. The accuracy and practicality of the model were verified by comparing it with various other models. The experimental results show that the usage of shared bicycles is higher during the peak commuting period and midday than during other periods; under the influence of specific special festivals and extreme weather, the usage of shared bicycles is significantly lower. Combined with the experimental results, it can provide scientific and practical guidance for managing and optimizing bike sharing placement and scheduling.