Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Bikelution: Federated Gradient-Boosting for Scalable Shared Micro-Mobility Demand Forecasting

Antonios Tziorvas, Andreas Tritsarolis, Yannis Theodoridis

TL;DR

Bikelution is proposed, an efficient Federated Learning (FL) solution based on gradient-boosted trees that preserves privacy while delivering accurate mid-term demand forecasts up to six hours ahead that highlights the feasibility of privacy-aware demand forecasting and outlines the trade-offs between FL and CML approaches.

Abstract

The rapid growth of dockless bike-sharing systems has generated massive spatio-temporal datasets useful for fleet allocation, congestion reduction, and sustainable mobility. Bike demand, however, depends on several external factors, making traditional time-series models insufficient. Centralized Machine Learning (CML) yields high-accuracy forecasts but raises privacy and bandwidth issues when data are distributed across edge devices. To overcome these limitations, we propose Bikelution, an efficient Federated Learning (FL) solution based on gradient-boosted trees that preserves privacy while delivering accurate mid-term demand forecasts up to six hours ahead. Experiments on three real-world BSS datasets show that Bikelution is comparable to its CML-based variant and outperforms the current state-of-the-art. The results highlight the feasibility of privacy-aware demand forecasting and outline the trade-offs between FL and CML approaches.

Bikelution: Federated Gradient-Boosting for Scalable Shared Micro-Mobility Demand Forecasting

TL;DR

Bikelution is proposed, an efficient Federated Learning (FL) solution based on gradient-boosted trees that preserves privacy while delivering accurate mid-term demand forecasts up to six hours ahead that highlights the feasibility of privacy-aware demand forecasting and outlines the trade-offs between FL and CML approaches.

Abstract

The rapid growth of dockless bike-sharing systems has generated massive spatio-temporal datasets useful for fleet allocation, congestion reduction, and sustainable mobility. Bike demand, however, depends on several external factors, making traditional time-series models insufficient. Centralized Machine Learning (CML) yields high-accuracy forecasts but raises privacy and bandwidth issues when data are distributed across edge devices. To overcome these limitations, we propose Bikelution, an efficient Federated Learning (FL) solution based on gradient-boosted trees that preserves privacy while delivering accurate mid-term demand forecasts up to six hours ahead. Experiments on three real-world BSS datasets show that Bikelution is comparable to its CML-based variant and outperforms the current state-of-the-art. The results highlight the feasibility of privacy-aware demand forecasting and outline the trade-offs between FL and CML approaches.
Paper Structure (9 sections, 6 equations, 3 figures, 2 tables)

This paper contains 9 sections, 6 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Formulation and Proposed Methodology
  4. Problem Definition
  5. Extending to Federated Learning
  6. Experimental Study
  7. Experimental Setup, Datasets and Preprocessing
  8. Experimental Results
  9. Conclusions

Figures (3)

  • Figure 1: Overview of our Bikelution forecasting framework.
  • Figure 2: Bikelution Federated Workflow Overview
  • Figure 3: Bike demand forecasting (arrivals) at three representative stations on the test set of the NYC dataset during March 17, 2022. Each row corresponds to bike stations whose HFL-global RMSE falls at the 25th (top), 50th (middle), and 75th (bottom) percentiles. Each column corresponds to one of the six prediction horizons (left-to-right: 1-6 hours). Solid blue lines represent the actual demand, while the green and orange lines show the predictions of the CML and HFL-global models, respectively.

Theorems & Definitions (3)

  • Definition 1: Bike trip
  • Definition 2: Bike demand
  • Definition 3: Bike demand forecasting