Deep Learning for Sector-Specific Labor Market Forecasting: Integrating Job Postings and Macroeconomic Indicators
This paper presents a sector-specific employment forecasting framework that integrates deep learning with heterogeneous labor market data, including job postings and macroeconomic indicators. Traditional statistical methods often fail to capture the non-linear, high-dimensional dynamics of modern la...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11086536/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This paper presents a sector-specific employment forecasting framework that integrates deep learning with heterogeneous labor market data, including job postings and macroeconomic indicators. Traditional statistical methods often fail to capture the non-linear, high-dimensional dynamics of modern labor markets, especially under volatile economic conditions. To address this, we propose the Stochastic Adaptive Labor Integration Network (SALIN) and the Hierarchical Opportunity-Aware Mechanism (HOAM). SALIN models the labor market as a stochastic bipartite graph enhanced with latent behavioral embeddings and temporal learning components, enabling fine-grained simulation of adaptive interactions between job seekers and firms. HOAM supports multi-level policy interventions through fairness-aware optimization and dynamic opportunity reallocation. Together, these modules facilitate short-to-medium term forecasting at both national and regional levels, offering decision support for policymakers, economic analysts, and labor market platforms. Empirical evaluations across diverse datasets demonstrate improved accuracy and robustness in forecasting employment trends, particularly in sector-sensitive and disruption-prone contexts. Empirical results show that our framework reduces MAE by up to 11.2% and MAPE by up to 10.6% compared to state-of-the-art baselines such as FEDformer, across diverse real-world labor datasets. |
|---|---|
| ISSN: | 2169-3536 |