HouseTS: A Large-Scale, Multimodal Spatiotemporal U.S. Housing Dataset

2025-06-01

AI Summary

Existing long-term housing price forecasting lacks a large-scale, reproducible benchmark dataset that simultaneously captures spatiotemporal depth, multimodal richness, and real-world granularity. Method: We construct a monthly multimodal spatiotemporal dataset covering 6,000 ZIP codes across 30 U.S. metropolitan areas from 2012–2023 (890K+ records), integrating satellite imagery, point-of-interest (POI) data, socioeconomic indicators, fine-grained real-estate features, and natural-language descriptions. We propose a novel multimodal urban evolution analysis paradigm, unifying vision-language models (VLMs), POI embeddings, geographically weighted feature engineering, and a unified time-series modeling framework. Contribution/Results: We release an open-source dataset and a standardized benchmark comprising 14 baseline models. Our approach enables interpretable, satellite-driven textual generation of urban change narratives, significantly enhancing prediction transparency and policy-relevance. The benchmark establishes new standards for reproducibility, scalability, and multimodal integration in urban forecasting.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Multi-modal Vision

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Accurate house-price forecasting is essential for investors, planners, and researchers. However, reproducible benchmarks with sufficient spatiotemporal depth and contextual richness for long horizon prediction remain scarce. To address this, we introduce HouseTS a large scale, multimodal dataset covering monthly house prices from March 2012 to December 2023 across 6,000 ZIP codes in 30 major U.S. metropolitan areas. The dataset includes over 890K records, enriched with points of Interest (POI), socioeconomic indicators, and detailed real estate metrics. To establish standardized performance baselines, we evaluate 14 models, spanning classical statistical approaches, deep neural networks (DNNs), and pretrained time-series foundation models. We further demonstrate the value of HouseTS in a multimodal case study, where a vision language model extracts structured textual descriptions of geographic change from time stamped satellite imagery. This enables interpretable, grounded insights into urban evolution. HouseTS is hosted on Kaggle, while all preprocessing pipelines, benchmark code, and documentation are openly maintained on GitHub to ensure full reproducibility and easy adoption.

Problem

• Lack of large-scale housing datasets with spatiotemporal depth
• Need for standardized benchmarks in house-price forecasting
• Limited multimodal integration of housing data with contextual factors

Innovation

• Large-scale multimodal housing dataset
• Standardized benchmarks with 14 models
• Vision language model for urban insights