Electric Kickboard Demand Prediction in Spatiotemporal Dimension Using Clustering-Aided Bagging Regressor

Abstract

Demand for electric kickboards is increasing specifically in tourist-centric regions worldwide. In order to gain a competitive edge and to provide quality service to customers, it is essential to properly deploy rental electric kickboards (e-kickboards) at the time and place customers want. However, it is necessary to study how to divide the region to predict electric mobility demand by region. Therefore, this study is made to more accurately predict future demand based on past regional customers’ electric mobility demand data. We have proposed a novel electric kickboard demand prediction in spatiotemporal dimension using clustering-aided bagging regressor. We have used electric kickboard usage data from a Jeju, South Korea-based company. As a result of the experiment, it was found that the accuracy before using clustering-based bagging regressor and when the region was divided by the clustering method, the performance was improved, and we have achieved a regression score R 2 of 93.42 using our proposed approach. We have compared our proposed approach with other state-of-the-art models, and we have also compared our model with different other combinations of bagging regressors. This study can be helpful for companies to meet the user’s demand for a better quality of service.

Stacking Machine Learning Models Empowered High Time-Height-Resolved Ozone Profiling from the Ground to the Stratopause Based on MAX-DOAS Observation

“…ML models were employed to regress these feature variables against O 3 concentrations at each altitude layer using 5-fold cross-validation. Four ML models were selected as the base-layer models, namely, BaggingRegressor (BR), 27 GradientBoostingRegressor (GBR), 28 XGBRegressor (XGBR), 29 and LGBMRegressor (LGBR). 30 MutipleLinearRegressor (MLR), as the meta-model, is used to stack these ML models for integrating the collective knowledge gained from the base models' predictions.…”

Optimizing Shared E-Scooter Operations Under Demand Uncertainty: A Framework Integrating Machine Learning and Optimization Techniques

“…In this hybrid case, the ILP solver optimizes the pickups and drop-offs for a known route sequence (… → 𝑖 → 𝑗 → ⋯, ∀𝑖, 𝑗𝜖𝑁 ) given by the ACO, and the ACO then combines the penalty cost with the driving distance cost to improve the route sequence iteratively. Therefore, the ILP formulation of rebalancing for a known route sequence is Here, Constraints ( 28)-( 31) are pickup and drop-off constraints at the current node 𝑗. Equations ( 32), (34), and (36) give the accumulated numbers of faulty, low-battery, and usable e-scooters, respectively, that are on the rebalancing vehicle at the current node 𝑗, defined as the sum of those that are on the rebalancing vehicle at the previous node 𝑖 and the number picked up or dropped off at the current node 𝑗. Equations ( 33) and (35) give the remaining numbers of faulty and low-battery e-scooters, respectively. Equations ( 37) and (38) give the unmet demands and excessive numbers of escooters, respectively.…”

“…True demand data is highly desirable for operational planning; however, it is often inaccessible unless operators grant access to extract it from user app activities [30]. Due to data limitations, historical ridership data are commonly used to evaluate the effectiveness of proposed models or frameworks [4,28,29,33,34,70]. Similarly, this study employs historical data as a case study, while the potential demand is managed through the safety stock parameter or minimum number of usable e-scooters (𝐶 𝑖 ).…”

“…Furthermore, a long short-term memory (LSTM) was used to forecast the hourly demand of shared e-scooters in the Seocho and Gangnam districts of Seoul, South Korea [33]. More recently, Khan et al [34] devised a bagging ensemble model based on extreme gradient boosting (XGBoost), extra trees, and random forests to forecast the (k-means algorithm-)clustered daily demand of shared escooters deployed on Jeju Island, South Korea.…”

Prediction of Corrosion Inhibition Efficiency Based on Machine Learning for Pyrimidine Compounds: A Comparative Study of Linear and Non-linear Algorithms

“…The Bagging Regressor, which is a non-linear model algorithm, has yielded satisfactory predictive results [16,17]. The objective of this study is to assess and compare the effectiveness of linear and non-linear algorithm models in a machine learning-based quantitative structure-property relationship (QSPR) approach.…”