Analysis Of The Spatio-Temporal Patterns Of Shared Bikes From City-Block Perspective

Precise and detailed identification for urban functions is the basis for perfecting land use policy and land use development,especially for the high-quality protection and renewal of the Beijing core area.Street blocks are carriers of residents’ daily life.Affected by the functional differentiation of urban areas,residents’ travel in each block will show specific patterns,which help to identify the functional type of the block quantitatively.However,there is few works concentrating on the topic of the functional types recognition of city blocks with shared bikes data in the literature,and due to the various spatial sizes of blocks,it is difficult to directly reflect the spatio-temporal patterns of residents’ mobility at the block scale.Shared bikes data has advantages in detailed data volume,high positioning accuracy and fair real-time performance.Therefore,we proposed a city block type identification framework based on the spatiotemporal pattern interpretation of shared bikes,including block-scale shared bikes flow modeling,block-scaled difference calculation,and block type identification based on clustering "shared bikes intensity spectrums".This work provides a new insight for the analysis of spatio-temporal patterns of shared bikes and the identification of block types.The main contents of this thesis are as follows:(1)To accurately extract the characteristics of shared bikes data and obtain the spatio-temporal patterns at the block scale,we established a workflow to preprocess and analyze shared bikes data.Results showed that using intra-and inter-block bicycle flow characteristics could provide effective information for discerning differences between blocks.(2)Focusing on the modeling of bicycle flow characteristics at the block scale,we developed a method for modeling the flow characteristics of shared bicycles based on block distance.Three quantitative indicators are specifically designed to measure the bicycle flow within the block,between the blocks and from temporal aspect.On this basis,the cosine similarity method is introduced to evaluate the block difference.Our results indicated that the constructed shared bikes flow features have strong discrimination ability,which lays the foundation for realizing identification of block types.(3)Aiming at the flow characteristics of shared bikes with high data dimension and large redundancy.The part introduced sparse subspace clustering to cluster blocks by constructing "shared bikes intensity spectrums",and tried to distinguish and separate mixed blocks.We got two types of single function blocks and three types of mixed function blocks,and accuracy was evaluated by confusion matrix,with the overall accuracy being 71.52%,kappa coefficient being 0.614.The above analysis verified the effectiveness of applying shared bikes data to explore the spatio-temporal patterns of residents’ mobility in the blocks of Beijing core area,and it is effective to extract functional features of blocks and identify precise block types.