Publicación: Análisis predictivo y de Big Data para la comprensión del comportamiento del consumidor en el sector inmobiliario del Valle de Aburrá
Portada
Citas bibliográficas
Código QR
Director
Autor corporativo
Recolector de datos
Otros/Desconocido
Director audiovisual
Editor/Compilador
Editores
Tipo de Material
Fecha
Cita bibliográfica
Título de serie/ reporte/ volumen/ colección
Es Parte de
Resumen
RESUMEN: El trabajo titulado “Análisis Predictivo y de Big Data para la comprensión del comportamiento del consumidor en el sector inmobiliario del Valle de Aburrá” explora cómo las empresas inmobiliarias pueden utilizar técnicas avanzadas de análisis de datos para anticipar necesidades y preferencias de los clientes. A través del uso de fuentes como redes sociales, portales web, registros de ventas, variables demográficas y datos de sensores urbanos, se aplicaron modelos predictivos como Random Forest, XGBoost y Redes Neuronales, siendo estas últimas las más precisas (R² = 0,93; MAE = 24,9 millones COP). Se definieron cinco criterios de evaluación para seleccionar modelos según el nivel de madurez tecnológica de la empresa. Los resultados permiten diseñar estrategias de marketing personalizadas basadas en datos, validadas mediante focus groups que resaltaron el interés por soluciones ajustadas a presupuesto, ubicación y estilo de vida, promoviendo así una toma de decisiones más efectiva e innovadora en el sector inmobiliario.
Resumen en inglés
ABSTRACT: The project entitled “Predictive and Big Data Analysis for Understanding Consumer Be- havior in the Real Estate Sector of the Aburrá Valley” has as its main objective to explore how real estate companies can apply advanced data analysis techniques to anticipate the needs and preferences of their customers. In a highly competitive and constantly changing environment, this study seeks to provide tools and strategies that enable companies to better align their offerings with actual market demand. Through a methodology based on the collection and analysis of Big Data, the project iden- tifies the main sources of information available in the real estate sector, such as social media, sales records, and demographic data. In addition, different predictive modeling techniques, such as ma- chine learning, are evaluated to determine which ones offer the best results in predicting consumer trends and behaviors. The expected results include a set of personalized, data-driven marketing strategies that will improve the customer experience and optimize sales operations in companies in the sector. At the institutional level, the project encourages collaboration between universities, companies, and government entities, promoting the adoption of innovative, data-driven solutions in the real estate sector. The main results of the study include the identification and prioritization of five key Big Data sources (web portals, social networks, sales records, demographic variables, and urban sensor data), highlighted for their availability and relevance in the Aburrá Valley. In the simulated hous- ing price prediction pilot tests, Random Forest achieved an R² of 0.88 with an approximate MAE of 29.4 million COP; XGBoost showed an R² of 0.91 and an MAE of 26.8 million COP; while Neural Networks achieved the best performance with an R² of 0.93 and an MAE of 24.9 million COP. In addition, five evaluation criteria were established—accuracy, interpretability, technical requirements, response time, and scalability—to select the most appropriate technique according to the company's level of technological maturity, recommending Random Forest for initial imple- mentations, XGBoost in intermediate phases, and Neural Networks when maximum accuracy is the priority. Finally, the focus groups validated a clear demand for more personalized marketing experiences, with preferences for alerts and recommendations tailored to budget, location, and lifestyle.