Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DeepSTA: A Spatial-Temporal Attention Network for Logistics Delivery Timely Rate Prediction in Anomaly Conditions

Jinhui Yi, Huan Yan, Haotian Wang, Jian Yuan, Yong Li

TL;DR

This work targets predicting couriers' delivery timely rates in anomaly conditions, such as epidemic outbreaks, where external factors and sparse anomalous data hinder standard forecasting. It introduces DeepSTA, a two-module framework that combines a spatio-temporal learning module (road districts via Node2vec, couriers via GCN, temporal dynamics via LSTM, and external anomaly signals via RNN) with a memory-based anomaly pattern attention module to store and retrieve abnormal-event patterns. Empirical results on real JD Logistics data show DeepSTA surpasses baselines by significant margins (MAE: ~12%, MSE: ~14%), and ablation studies confirm the contribution of each component. The model is deployed online at JD Logistics, highlighting practical impact for mitigating anomaly-induced delivery delays in logistics networks.

Abstract

Prediction of couriers' delivery timely rates in advance is essential to the logistics industry, enabling companies to take preemptive measures to ensure the normal operation of delivery services. This becomes even more critical during anomaly conditions like the epidemic outbreak, during which couriers' delivery timely rate will decline markedly and fluctuates significantly. Existing studies pay less attention to the logistics scenario. Moreover, many works focusing on prediction tasks in anomaly scenarios fail to explicitly model abnormal events, e.g., treating external factors equally with other features, resulting in great information loss. Further, since some anomalous events occur infrequently, traditional data-driven methods perform poorly in these scenarios. To deal with them, we propose a deep spatial-temporal attention model, named DeepSTA. To be specific, to avoid information loss, we design an anomaly spatio-temporal learning module that employs a recurrent neural network to model incident information. Additionally, we utilize Node2vec to model correlations between road districts, and adopt graph neural networks and long short-term memory to capture the spatial-temporal dependencies of couriers. To tackle the issue of insufficient training data in abnormal circumstances, we propose an anomaly pattern attention module that adopts a memory network for couriers' anomaly feature patterns storage via attention mechanisms. The experiments on real-world logistics datasets during the COVID-19 outbreak in 2022 show the model outperforms the best baselines by 12.11% in MAE and 13.71% in MSE, demonstrating its superior performance over multiple competitive baselines.

DeepSTA: A Spatial-Temporal Attention Network for Logistics Delivery Timely Rate Prediction in Anomaly Conditions

TL;DR

This work targets predicting couriers' delivery timely rates in anomaly conditions, such as epidemic outbreaks, where external factors and sparse anomalous data hinder standard forecasting. It introduces DeepSTA, a two-module framework that combines a spatio-temporal learning module (road districts via Node2vec, couriers via GCN, temporal dynamics via LSTM, and external anomaly signals via RNN) with a memory-based anomaly pattern attention module to store and retrieve abnormal-event patterns. Empirical results on real JD Logistics data show DeepSTA surpasses baselines by significant margins (MAE: ~12%, MSE: ~14%), and ablation studies confirm the contribution of each component. The model is deployed online at JD Logistics, highlighting practical impact for mitigating anomaly-induced delivery delays in logistics networks.

Abstract

Prediction of couriers' delivery timely rates in advance is essential to the logistics industry, enabling companies to take preemptive measures to ensure the normal operation of delivery services. This becomes even more critical during anomaly conditions like the epidemic outbreak, during which couriers' delivery timely rate will decline markedly and fluctuates significantly. Existing studies pay less attention to the logistics scenario. Moreover, many works focusing on prediction tasks in anomaly scenarios fail to explicitly model abnormal events, e.g., treating external factors equally with other features, resulting in great information loss. Further, since some anomalous events occur infrequently, traditional data-driven methods perform poorly in these scenarios. To deal with them, we propose a deep spatial-temporal attention model, named DeepSTA. To be specific, to avoid information loss, we design an anomaly spatio-temporal learning module that employs a recurrent neural network to model incident information. Additionally, we utilize Node2vec to model correlations between road districts, and adopt graph neural networks and long short-term memory to capture the spatial-temporal dependencies of couriers. To tackle the issue of insufficient training data in abnormal circumstances, we propose an anomaly pattern attention module that adopts a memory network for couriers' anomaly feature patterns storage via attention mechanisms. The experiments on real-world logistics datasets during the COVID-19 outbreak in 2022 show the model outperforms the best baselines by 12.11% in MAE and 13.71% in MSE, demonstrating its superior performance over multiple competitive baselines.
Paper Structure (27 sections, 5 equations, 5 figures, 3 tables)

This paper contains 27 sections, 5 equations, 5 figures, 3 tables.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORKS
  3. Online Business Prediction Problem
  4. Time Series Prediction Methods
  5. Anomaly Learning
  6. PROBLEM FORMULATION
  7. METHOD
  8. The Proposed Framework
  9. Anomaly Spatio-Temporal Learning Module
  10. Road District Embedding Component
  11. Courier Spatial Learning Component
  12. Courier Temporal Learning Component
  13. Anomaly Learning Component
  14. Anomaly Pattern Attention Module
  15. Model Training
  16. ...and 12 more sections

Figures (5)

  • Figure 1: (a) District A is far from B and C which are adjacent. In May 2022, the timely rate on B and C dropped synchronously due to the regional epidemic lockdown. (b) On October 28th, courier B was absent due to the lockdown, and his parcels were reassigned to A and C.
  • Figure 2: Overall architecture of DeepSTA.
  • Figure 3: Visualization of external factors.
  • Figure 4: The impact of hyper-parameters.
  • Figure 5: A case study of deployed models.