Propuesta de método de estimación de rentabilidad en proyectos de la industria inmobiliaria utilizando tecnologías de machine learning.
Loading...
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
El presente informe destaca la evolución histórica de la evaluación de proyectos inmobiliarios. Se aborda la creciente importancia de considerar el riesgo en las evaluaciones financieras, recalcando la necesidad de adaptación a medida que los proyectos evolucionan en el tiempo.
La digitalización ha revolucionado el sector inmobiliario, simplificando transacciones y mejorando la experiencia del cliente, acentuando las tendencias digitales que la industria inmobiliaria ha seguido en los últimos tiempos. La inteligencia artificial se ha vuelto crucial en la toma de decisiones financieras, permitiendo una gestión más precisa y la predicción de eventos clave mediante el uso de herramientas de machine learning, resaltando su potencial en la evaluación integral de proyectos inmobiliarios, ofreciendo la capacidad de identificar tendencias y patrones para mejorar la rentabilidad. El uso de modelos de machine learning logra una mejor predicción de la rentabilidad final de los proyectos inmobiliarios. La selección de variables que más influyen en la rentabilidad de cada inmobiliaria en conjunto con la elección del modelo de machine learning a utilizar, juegan un papel fundamental en la implementación de inteligencia artificial para evaluaciones financieras de este tipo de proyectos.
Para datos simulados tratados en la presente memoria de título, los modelos de machine learning llamados Linear Regression y Linear Support Vector Regressor son los que demuestran ser más adecuados (dentro de un total de 5 modelos) para predecir la rentabilidad final de proyectos inmobiliarios de acuerdo con los criterios de, mean square error y coefficient of determination, además de un análisis de incertidumbre basado tanto en las correlaciones de las variables usadas, como también de una evaluación k-fold para cada modelo.
This report highlights the historical evolution of real estate project evaluation. It addresses the growing importance of considering risk in financial evaluations, emphasizing the need for adaptation as projects evolve over time. Digitization has revolutionized the real estate sector, simplifying transactions and improving customer experience, accentuating the digital trends that the real estate industry has followed in recent times. Artificial intelligence has become crucial in financial decision-making, allowing for more precise management and the prediction of key events through the use of machine learning tools, highlighting its potential in the comprehensive evaluation of real estate projects, offering the ability to identify trends and patterns to improve profitability. The use of machine learning models achieves a better prediction of the final profitability of real estate projects. The selection of variables that most influence the profitability of each real estate agency together with the choice of the machine learning model to be used, play a fundamental role in the implementation of artificial intelligence for financial evaluations of this type of projects. For simulated data treated in this thesis, the machine learning models called Linear Regression and Linear Support Vector Regressor are those that prove to be the most suitable (out of a total of 5 models) to predict the final profitability of real estate projects according to the criteria of mean square error and coefficient of determination, in addition to an uncertainty analysis based on both the correlations of the variables used, as well as a k-fold evaluation for each model.
This report highlights the historical evolution of real estate project evaluation. It addresses the growing importance of considering risk in financial evaluations, emphasizing the need for adaptation as projects evolve over time. Digitization has revolutionized the real estate sector, simplifying transactions and improving customer experience, accentuating the digital trends that the real estate industry has followed in recent times. Artificial intelligence has become crucial in financial decision-making, allowing for more precise management and the prediction of key events through the use of machine learning tools, highlighting its potential in the comprehensive evaluation of real estate projects, offering the ability to identify trends and patterns to improve profitability. The use of machine learning models achieves a better prediction of the final profitability of real estate projects. The selection of variables that most influence the profitability of each real estate agency together with the choice of the machine learning model to be used, play a fundamental role in the implementation of artificial intelligence for financial evaluations of this type of projects. For simulated data treated in this thesis, the machine learning models called Linear Regression and Linear Support Vector Regressor are those that prove to be the most suitable (out of a total of 5 models) to predict the final profitability of real estate projects according to the criteria of mean square error and coefficient of determination, in addition to an uncertainty analysis based on both the correlations of the variables used, as well as a k-fold evaluation for each model.
Description
Tesis presentada para optar al título de Ingeniero Civil Industrial
Keywords
Rentabilidad Evaluación, Comercio inmobiliario, Aprendizaje de máquina