Analysis of Shared Bicycle Usage based on K-Means and GMM Clustering Algorithm

Bicycle sharing in cities facilitated people’s daily travel. The convenience it offers made the sharing model a success in the beginning. However, as the number of users increases, assigning the shared bicycles reasonably to maximize their availability becomes a challenge for suppliers. Classifying the users and predicting their demands are key to solving the problem. The author used historical data to visualize the relationship between travel time and bicycle usage and understand the users’ travel patterns. Principal Component Analysis(PCA) was performed to reduce the dimensionality of the data, and the k-means and Gaussian mixture model (GMM) clustering methods were used to classify the users. In this case, k-means clustering did not effectively classify the data, while GMM clustering divided the data into two categories. With GMM, the users’ travel patterns were associated with weekdays. This article also explores the pros and cons of K-means and GMM are suitable for continuous and discontinuous data through specific cases.