Sub-City Real Estate Price Index Forecasting at Weekly Horizons Using Satellite Radar and News Sentiment

Paper: arXiv 2602.18572

Authors: Baris Arat, Hasan Fehmi Ates, Emre Sefer

Abstract

Reliable real estate price indicators are typically published at city level and low frequency, limiting their use for neighborhood-scale monitoring and long-horizon planning. We study whether sub-city price indices can be forecasted at weekly frequency by combining physical development signals from satellite radar with market narratives from news text. Using over 350,000 transactions from Dubai Land Department (2015-2025), we construct weekly price indices for 19 sub-city regions and evaluate forecasts from 2 to 34 weeks ahead. Our framework fuses regional transaction history with Sentinel-1 SAR backscatter, news sentiment combining lexical tone and semantic embeddings, and macroeconomic context. Results are strongly horizon dependent: at horizons up to 10 weeks, price history alone matches multimodal configurations, but beyond 14 weeks sentiment and SAR become critical. At long horizons (26-34 weeks), the full multimodal model reduces mean absolute error from 4.48 to 2.93 (35% reduction), with gains statistically significant across regions. Nonparametric learners consistently outperform deep architectures in this data regime. These findings establish benchmarks for weekly sub-city index forecasting and demonstrate that remote sensing and news sentiment materially improve predictability at strategically relevant horizons.

Complexity vs Empirical Score

• Math Complexity: 3.5/10
• Empirical Rigor: 8.5/10
• Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper presents a highly rigorous empirical framework for forecasting sub-city real estate price indices using a multi-modal approach combining satellite SAR imagery and news sentiment. While the mathematical formulations are straightforward standard machine learning and time-series feature constructions, the dataset scale (350k+ transactions over 10 years), temporal granularity (weekly horizon), and comprehensive ablation studies are outstanding.

Research Flowchart

  flowchart TD
    A[Research Goal: Forecast Weekly Sub-City Real Estate Price Indices at 2 to 34 Week Horizons] --> B[Multi-Source Data Ingestion]
    
    B -->|350k+ Dubai Transactions| C1[19 Regional Price Indices]
    B -->|Sentinel-1 SAR Radar| C2[Physical Development Signals]
    B -->|News Text & Lexicon/Embeddings| C3[Market Sentiment Narratives]
    
    C1 & C2 & C3 --> D[Multimodal Feature Fusion & ML Model Training]
    
    D --> E{Horizon-Dependent Forecasting Performance}
    
    E -->|Short Horizon: <= 10 Weeks| F[Historical Price Alone is Sufficient]
    E -->|Long Horizon: 26 to 34 Weeks| G[Full Multimodal Model Yields 35% MAE Reduction]

Cited References (56)

1. MugRep : A Multi - Task Hierarchical Graph Representation Learning Framework for Real Estate Appraisal (2021) — link
2. Financial asset price prediction with graph neural network-based temporal deep learning models (2025) — link
3. Multi-Context Real Estate Market Prediction (2025) — link
4. Sentence- BERT : Sentence Embeddings using Siamese BERT -Networks (2019) — link
5. Interest rates and real estate prices: a panel study (2023)
6. Experimental research on the impact of interest rate on real estate market transactions (2022)
7. Construction and Application of Property Price Indices (2018) — link
8. Construction of property price indices: temporal aggregation and accuracy of various index methods (2013)

0 of 56 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.