Land Use Identification of the Metropolitan Area of Guadalajara Using Bicycle Data: an Unsupervised Classification Approach

This paper presents the results from a research project that is focused on land-use, land-mapping and human behavior analysis to evaluate its movement through information sources that contains geo-referenced information. The source used is data provided by MiBici, which is a bicycle sharing platform established in the city of Guadalajara (state of Jalisco) in Mexico. The database includes a combination of different data for every month of recent years, and the access to this information is free and totally available. The methodologies used for this research were agile development techniques for project planning, k-Nearest Neighbors, Decision Trees, and K-Means for the clusterization of different zones. The programming language used for the development was Python and in addition an implementation proposal was performed using the Amazon Web Services platform. The results obtained shows that the different clusters obtained are well defined, the data is classified among them and physically the information have a valuable relationship to the human activity and behavior within the Metropolitan Area of Guadalajara.