Liu, Ziwen;
Orr, Scott;
Grau-Bove, Josep;
(2023)
HTGTM: Hybrid Temporal-Graph Tabular Model for Complex Multimodal Tabular Data Processing.
In:
Proceedings IEEE 33rd International Workshop on Machine Learning for Signal Processing.
IEEE (Institute of Electrical and Electronics Engineers): Rome,Italy.
(In press).
Preview |
Text
HTGTM.pdf - Accepted Version Download (444kB) | Preview |
Abstract
Understanding the behaviour of volunteers is an important research area in social science and management. This paper tackles the Volunteer Retention Prediction task of the IEEE MLSP 2023 Data Challenge, utilizing a dataset from the COVID-19 pandemic volunteer coordination in Shenzhen, China in 2020 ‒ 2021, with the objective of forecasting volunteer retention in the next ten months. Our paper proposes Hybrid Temporal-Graph Tabular Model (HTGTM), a deep learning-based hybrid model designed to extract and analyze temporal and graph information within complex tabular data. In this data challenge, we compared our model, ensembled with XGBoost, against traditional machine learning methods and deep learning models that are specifically tailored for tabular data. Our method exhibited robust performance, validated by its 1st-ranking Root-Mean-Square-Error (RMSE) score of 76.36 in MLSP 2023 Data Challenge Kaggle private Leaderboard. This research sheds light on pertinent volunteer retention prediction tasks and highlights the incorporation of deep learning techniques in complex, multimodal tabular data processing tasks.
| Type: | Proceedings paper |
|---|---|
| Title: | HTGTM: Hybrid Temporal-Graph Tabular Model for Complex Multimodal Tabular Data Processing |
| Event: | 33rd IEEE International Workshop on Machine Learning for Signal Processing (MLSP 2023) |
| Location: | Rome, Italy |
| Dates: | 17 Sep 2023 - 20 Sep 2023 |
| ISBN-13: | 979-8-3503-2411-2 |
| Open access status: | An open access version is available from UCL Discovery |
| DOI: | 10.1109/MLSP55844.2023.10285918 |
| Publisher version: | https://doi.org/10.1109/MLSP55844.2023.10285918 |
| Language: | English |
| Additional information: | This version is the author accepted manuscript. For information on re-use, please refer to the publisher’s terms and conditions. |
| Keywords: | Volunteer Retention, Deep Learning, Multimodal Tabular Data, Ensemble Learning |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of the Built Environment UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of the Built Environment > Bartlett School Env, Energy and Resources |
| URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10174481 |
Archive Staff Only
 |
View Item |
