The number of rental bicycles has begun to increase worldwide. However, some problems followed, improper management and unreasonable promotion can lead to waste of resources and improper bicycle management. The current trend is that in the world, the problem of waste of resources exists in large numbers. The management department needs to assign and manage rental bicycles based on different data in different cities. We utilized the rental bicycle data collected in the University of California, Irvine Machine Learning Repository to score each data that affects the use of bicycles, and filter out variables suitable for analyzing the use of bicycles. After analysis, we found that eight variables are highly correlated with bicycle use. Considering that rental bicycles tend to cluster together, we analyzed the cluster characteristics of the bicycle dataset. The results show that the distribution of bicycles follows the cluster characteristics. Therefore, we designed a k-nearest neighbor (KNN) model to predict the traffic of rental bicycles. In order to verify the effectiveness of the scheme, we compared KNN with support vector regression, deep neural network, decision tree, and linear regression model. The experimental results show that our scheme can achieve the optimal prediction performance.
K-nearest Neighbors Regressor for Traffic Prediction of Rental Bikes
07.01.2022
1594509 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Improvement of Search Strategy With K-Nearest Neighbors Approach for Traffic State Prediction
Online Contents | 2016
|Improvement of Search Strategy With K-Nearest Neighbors Approach for Traffic State Prediction
Online Contents | 2015
|Green Urban Public Bikes Rental System: A Case Study of Taipei YouBike
Tema Archiv | 2013
|