Application of Density-Based Spatial Clustering of Application with Noise to Optimize Matching Problems in Ridesharing for Maximize Total Distance Proximity Index

Ridesharing is one of models that attempt to reduce congestion problems due to increased use of private vehicles with low occupancy. The problem related to ridesharing is to get an optimal pair of drivers and riders, while the numbers of participants involved are very large and optimization must be done in a short amount of time. In this paper DBSCAN clustering will be used as the first step to optimize the matching problem in ridesharing with DP Index as the objective function. Driver and rider with similar distance will be a good match by using DP Index as the objective function if both origin and destination of driver and rider are in a close proximity. DBSCAN clustering is one of the methods of clustering based on the density of 2-dimensional objects. In the initial stage, the DBSCAN clustering method is used to cluster the origin and destination locations of the drivers and riders. After obtaining the clusters, the driver-rider pair will be matched based on the maximum DP Index by the Hungarian algorithm. This paper uses three times periods as the result of this experiment shows that DBSCAN clustering able to increase the total number of pairs of driver-rider matching.