Mining Patterns Of Dockless Shared Bike Quantity Change In City

In recent years,the dockless shared bikes(DSB)have been experienced explosive growth in cities across the world.The emerging portable transportation method has also been used more and more widely in urban transportation.They are used as a convenient tool to connect,simplify or even replace some traditional modes of transportation.Shared bicycles largely distribute in bus stations,subway stations,business districts,residential areas and it is becoming an irreplaceable part of urban traffic and people’s daily life.However,the quick expansion of DSB has also bring several problems to the origin busy urban traffic.The main contradiction is that the supply of DSB may be unmatched with the need of DSB.For example,the users may find it is hard to find a bike nearby while in some other areas,many free bikes may be redundant.Some similar situations will also disturb the user’s travel plan,waste their time and even have harmful congestion in pedestrian walkways.Secondly,due to the frequency of urban activities and the strong mobility of people,operators need to have an optimized solution for redistributing the shared bicycles.By the explosive growth of DSB,the relative research becomes more and more popular.Scholars have studied different kinds of problems brought by DSB.Some of them study on the characteristics of shared bicycles.They mainly focus on the spatial distribution of shared bicycles in the city,how to allocate them at various sites to achieve cost optimization,and what the sharing bicycle system brings to urban life.Other papers discuss the implications of shared bicycle systems for urban transport,which focus on the impact of shared bicycle systems on urban roads and transportation planning.The main purpose of this paper is to explore the potential periodic characteristics from the spatial-temporal data of the urban shared bicycle system,and to make an abstract refinement of the distribution characteristics of shared bicycles in the city through a two-dimensional perspective.Therefore,it provides a new solution to solve the imbalance of supply demand and the redistribution of scheduling between regions.Compared with the previous researches on the characteristics of shared bicycles,our work is more challenging and innovative.How to describe and accurately define the general regularities contained in spatial-temporal data,and how to adopt a general method for exploring this kind of regularity across city are the difficulties which need to be solved in this paper.In response to these research objectives,we propose a concept called "Magnitude-Oblivious Periodical Patterns",which can be used to describe the regularity of dockless shared bike changes over time and across space,and this periodic pattern reflects the tendency of shared bicycle supply demand change over time,and it is also a means to help detect and balance the shared bicycle supply and demand in different urban areas.The advantage of using this pattern mining solution is that it can abandon the influence of scales between different regions,and essentially explore the common regularity of this periodic change.This research provides a ubiquitous approach rather than looking for a specific solution for a particular area.In view of the inherent periodic characteristics of urban shared bicycles in the region,we try to explore the periodic regularities existing in different regions,and explain the relationship between these periodic patterns,and explore the commonality and difference of periodic distribution between different regions.The main contributions of this paper are as follows:(1)We study the properties of shared bicycles in the “spatial-temporal”bi-dimension across the city,and propose a general concept to describe the periodic regularity in a certain geographical area,which called "Magnitude-Oblivious Periodical Patterns".(2)We learn various periodic detection methods to design an algorithm which fits the bicycle time-series data to detect its periodic property.Based on the periodicities,we further study how to use pattern mining algorithms to explore the periodic pattern of shared bicycles,and obtains a reasonable algorithm to find common regional periodic patterns in different areas.(3)We use the most widely covered dockless shared bike in China–Mobike as experiment.We obtained GPS data of Mobike in Beijing,Shanghai and Ningbo through crawling.And then we use the concepts and algorithms in(1)(2)to detect which areas have periodicity,and explore the periodic patterns in these cities.Experiments show that about 10%-13% of the geographical samples we selected have the property of periodicity.Meanwhile,we used Shanghai as an example to explore more than 150 periodical patterns.Furthermore,we analyze the rationality of these patterns and their practical significance through several study cases of periodic patterns.