Machine-Learning–Based Property Valuation for Dubai: Design, Implementation, and Evaluation

Yousif Alhammadi; Ahmad Ibrahim

Machine-Learning–Based Property Valuation for Dubai: Design, Implementation, and Evaluation

Yousif Alhammadi^*
, Ahmad Ibrahim

^*Corresponding author for this work

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

This paper presents a machine-learning framework for objective real-estate valuation in Dubai’s rapidly expanding property market. Using more than 700,000 transaction records from the Dubai Land Department, enriched with geospatial features, multiple machine-learning models were trained and evaluated. Rigorous preprocessing—including outlier removal, target encoding, and spatial feature engineering—significantly improved model stability. A Random Forest Regressor achieved the strongest performance, delivering a Mean Absolute Percentage Error (MAPE) of 11% and Mean Absolute Error (MAE) of approximately 1,144 AED. A full-stack system was developed to deploy the model for real-time valuation. Results indicate that machine-learning approaches can substantially outperform traditional valuation methods in accuracy and consistency.

Original language	English
Title of host publication	KSII The 17th International Conference on Internet (ICONI) 2025
Place of Publication	Okinawa, Japan
Publisher	Korean Society for Internet Information
Number of pages	2
Publication status	Published - 14 Dec 2025
Event	17th International Conference on Internet (ICONI 2025) - Okinawa, Japan, Okinawa, Japan Duration: 14 Dec 2025 → 17 Dec 2025 https://iconi.org/

Conference

Conference	17th International Conference on Internet (ICONI 2025)
Abbreviated title	ICONI 2025
Country/Territory	Japan
City	Okinawa
Period	14/12/25 → 17/12/25
Internet address	https://iconi.org/

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 15 Life on Land

Cite this

@inproceedings{28a6328272d244a1b79cf1a2d3a8fcbc,

title = "Machine-Learning–Based Property Valuation for Dubai: Design, Implementation, and Evaluation",

abstract = "This paper presents a machine-learning framework for objective real-estate valuation in Dubai{\textquoteright}s rapidly expanding property market. Using more than 700,000 transaction records from the Dubai Land Department, enriched with geospatial features, multiple machine-learning models were trained and evaluated. Rigorous preprocessing—including outlier removal, target encoding, and spatial feature engineering—significantly improved model stability. A Random Forest Regressor achieved the strongest performance, delivering a Mean Absolute Percentage Error (MAPE) of 11\% and Mean Absolute Error (MAE) of approximately 1,144 AED. A full-stack system was developed to deploy the model for real-time valuation. Results indicate that machine-learning approaches can substantially outperform traditional valuation methods in accuracy and consistency.",

author = "Yousif Alhammadi and Ahmad Ibrahim",

year = "2025",

month = dec,

day = "14",

language = "English",

booktitle = "KSII The 17th International Conference on Internet (ICONI) 2025",

publisher = "Korean Society for Internet Information",

note = "17th International Conference on Internet (ICONI 2025) , ICONI 2025 ; Conference date: 14-12-2025 Through 17-12-2025",

url = "https://iconi.org/",

}

TY - GEN

T1 - Machine-Learning–Based Property Valuation for Dubai: Design, Implementation, and Evaluation

AU - Alhammadi, Yousif

AU - Ibrahim, Ahmad

PY - 2025/12/14

Y1 - 2025/12/14

N2 - This paper presents a machine-learning framework for objective real-estate valuation in Dubai’s rapidly expanding property market. Using more than 700,000 transaction records from the Dubai Land Department, enriched with geospatial features, multiple machine-learning models were trained and evaluated. Rigorous preprocessing—including outlier removal, target encoding, and spatial feature engineering—significantly improved model stability. A Random Forest Regressor achieved the strongest performance, delivering a Mean Absolute Percentage Error (MAPE) of 11% and Mean Absolute Error (MAE) of approximately 1,144 AED. A full-stack system was developed to deploy the model for real-time valuation. Results indicate that machine-learning approaches can substantially outperform traditional valuation methods in accuracy and consistency.

AB - This paper presents a machine-learning framework for objective real-estate valuation in Dubai’s rapidly expanding property market. Using more than 700,000 transaction records from the Dubai Land Department, enriched with geospatial features, multiple machine-learning models were trained and evaluated. Rigorous preprocessing—including outlier removal, target encoding, and spatial feature engineering—significantly improved model stability. A Random Forest Regressor achieved the strongest performance, delivering a Mean Absolute Percentage Error (MAPE) of 11% and Mean Absolute Error (MAE) of approximately 1,144 AED. A full-stack system was developed to deploy the model for real-time valuation. Results indicate that machine-learning approaches can substantially outperform traditional valuation methods in accuracy and consistency.

M3 - Conference contribution

BT - KSII The 17th International Conference on Internet (ICONI) 2025

PB - Korean Society for Internet Information

CY - Okinawa, Japan

T2 - 17th International Conference on Internet (ICONI 2025)

Y2 - 14 December 2025 through 17 December 2025

ER -

Machine-Learning–Based Property Valuation for Dubai: Design, Implementation, and Evaluation

Abstract

Conference

UN SDGs

Fingerprint

Cite this