Research On Shared Bicycle Stock Prediction Based On Long-term And Short-term Memory Neural Network

The Bicycle Sharing System(BBS),especially the FPSS,provides great convenience for residents' lives.In recent years,the system has become more and more popular in cities,especially in the past two years.The city has begun to put this sharing model into use on a large scale.However,due to the lack of technology and experience of bicycle operators in the allocation of bicycle resources,various problems have arisen in the use of bicycles in different regions.What's more regrettable is that there are not many domestic studies on this issue.To this end,this article takes Shanghai as an example.Based on the latitude and longitude data of shared bicycles in the region,this paper attempts to put forward some suggestions for the allocation of bicycle resources.Specifically,the paper firstly analyzes the characteristics of shared bicycle traffic,and proposes the idea of regional clustering analysis of bicycle latitude and longitude data,and then using the time series forecasting model to predict the stock of bicycles under each region.After analyzing the collected data,the mean-shift clustering algorithm and the long-short-term memory neural network(LSTM)prediction algorithm were selected as the current use algorithm.The final paper is based on the comparison of the prediction error RMSE with the LSTM model and other models,and proves the applicability and efficiency of LSTM in predicting the inventory of bicycles.At the end of the text,the number of bicycle stocks in various regional points in Shanghai is also given for reference for subsequent bicycle rebalancing strategies.