AI-driven hybrid recommender model for personalized residential property selection in chennai based on flood risk, sustainability, and proximity
- 1Department of Civil Engineering, B.S. Abdur Rahman Crescent Institute of Science and Technology
Abstract
The real estate sector in Chennai is going through a dramatic change and home-buyers are looking for homes that embody their notions of safety, sustainability and access to services. However, traditional property search engines are unable to take into account all factors affecting the selection of properties, including flood and sustainability risk. The paper will be discussing a hybrid recommender system with the help of collaborative filtering, content-based filtering algorithms, ranking algorithms, reinforcement learning (RL) with the help of a Proximal Policy Optimization (PPO) algorithm. The input of the user in the organization is achieved by a standardized questionnaire, and real-time property data is used to provide customized property suggestions that are customized to the users’ needs dynamically. Some of the main challenges, including risk of flooding, sustainability and amenity and location factors, are being incorporated into the recommendations, adding context sensitivity and responsiveness to the environmental challenges. Data from 500 participants and 200 properties were used to validate the system. The system achieved 99.92% precision, 99.96% recall, 99.85% F1-score, and 85.2% NDCG@10. PPO enabled the model to dynamically adjust the collaborative filtering weight between 0.41 and 0.50 across 200 epochs based on user preferences. The results validate the proposed recommender system for dynamically suggesting available properties with customized recommendation and its flexibility and scalability compared to the traditional recommender systems. The research will pave the way towards improved property selection in flood-prone urban environment with climate consciousness and can be applied in other cities suffering from similar issues.
Related
- Residential proximity to oil and gas development and socioeconomic conditions: a comparative analysis of 33 countries
- Strategic green supplier selection to mitigate supply chain emissions: A Sustainability multidimensional approach
- A multimodal context-aware AI recommender for smart farming
- Online Measurements Reveal Contrasting Particle-Bound Reactive Oxygen Species and Chemical Composition in Emissions from Six Residential Solid Fuels
- Effects of anthropogenic heat on the urban wind-thermal environment in typical residential neighborhoods in Nanjing, China
BibTeX
@article{shiney2026aidriven,
title = {{AI-driven hybrid recommender model for personalized residential property selection in chennai based on flood risk, sustainability, and proximity}},
author = {G. Garmel Shiney and Dr K. Yogeswari},
journal = {Discover Sustainability},
year = {2026},
month = jul,
doi = {10.1007/s43621-026-04065-0},
url = {https://doi.org/10.1007/s43621-026-04065-0},
note = {CC BY},
}
RIS
TY - JOUR
TI - AI-driven hybrid recommender model for personalized residential property selection in chennai based on flood risk, sustainability, and proximity
AU - G. Garmel Shiney
AU - Dr K. Yogeswari
PY - 2026
DA - 2026/07/01
JO - Discover Sustainability
DO - 10.1007/s43621-026-04065-0
UR - https://doi.org/10.1007/s43621-026-04065-0
AB - The real estate sector in Chennai is going through a dramatic change and home-buyers are looking for homes that embody their notions of safety, sustainability and access to services. However, traditional property search engines are unable to take into account all factors affecting the selection of properties, including flood and sustainability risk. The paper will be discussing a hybrid recommender system with the help of collaborative filtering, content-based filtering algorithms, ranking algorithms, reinforcement learning (RL) with the help of a Proximal Policy Optimization (PPO) algorithm. The input of the user in the organization is achieved by a standardized questionnaire, and real-time property data is used to provide customized property suggestions that are customized to the users’ needs dynamically. Some of the main challenges, including risk of flooding, sustainability and amenity and location factors, are being incorporated into the recommendations, adding context sensitivity and responsiveness to the environmental challenges. Data from 500 participants and 200 properties were used to validate the system. The system achieved 99.92% precision, 99.96% recall, 99.85% F1-score, and 85.2% NDCG@10. PPO enabled the model to dynamically adjust the collaborative filtering weight between 0.41 and 0.50 across 200 epochs based on user preferences. The results validate the proposed recommender system for dynamically suggesting available properties with customized recommendation and its flexibility and scalability compared to the traditional recommender systems. The research will pave the way towards improved property selection in flood-prone urban environment with climate consciousness and can be applied in other cities suffering from similar issues.
SN - 2662-9984
ER -