Thesis Sistema de análisis espacial y predictivo para la caracterización del comercio informal en contextos urbanos
Loading...
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Telemática
Departament
Campus
Campus Santiago San Joaquín
Abstract
El comercio informal en ámbitos urbanos plantea retos significativos para la gestión del espacio público y la formulación de decisiones fundamentadas en evidencia. Con el objetivo de abordar este fenómeno en el centro de Santiago, este estudio desarrolla una plataforma integral que integra técnicas de inteligencia artificial (incluyendo visionado automatizado mediante YOLO y clasificación probabilística con Naive Bayes), análisis geoespacial (empleando una grilla de 100x100 metros) y el procesamiento de imágenes georreferenciadas. Esta solución posibilita caracterizar y analizar la ocupación de comerciantes en la vía pública, identificar patrones, estimar probabilidades de presencia, y generar visualizaciones interactivas. A pesar de que las primeras evaluaciones se llevaron a cabo con datos y etiquetas simuladas, la validación cruzada avala la correcta implementación del flujo de trabajo, estableciendo las bases para la integración de datos reales, la adopción de técnicas de aprendizaje profundo y la aplicación del enfoque a otros contextos urbanos. De esta manera, se pretende contribuir a un equilibrio entre el rol económico-social del comercio informal y la ordenación del espacio público. Para concretar esta propuesta, se consideran los siguientes pasos principales:
1. Diseñar y desarrollar un prototipo de interfaz que consolide y exhiba la información recolectada, resaltando las áreas con mayor ocupación y sus atributos más relevantes.
2. Implementar un mecanismo de carga de datos desde múltiples fuentes (imágenes, registros oficiales), permitiendo su procesamiento con algoritmos de visión por computador y análisis geoespacial.
3. Crear visualizaciones de resultados, incorporando gráficos y mapas para evidenciar patrones y tendencias en la distribución del comercio informal.
4. Integrar métodos de aprendizaje automático que faciliten la clasificación y segmentación de la información, favoreciendo el uso de evidencia en la toma de decisiones.
5. Realizar pruebas de validación para asegurar la precisión y consistencia de los algoritmos, evaluando su eficiencia al identificar y analizar los patrones de comercio informal.
Este enfoque multidisciplinario, pionero en Chile, profundiza el entendimiento de un fenómeno que afecta el desarrollo económico-social y la planificación urbana, ofreciendo una herramienta sólida para la gestión del comercio informal y el diseño de políticas públicas basadas en evidencia.
Informal trade within urban settings presents significant challenges to the management of public spaces and the formulation of decisions grounded in empirical evidence. In response to this issue in the context of downtown Santiago, this study develops a comprehensive platform integrating artificial intelligence methodologies, specifically automated vision via YOLO and probabilistic classification using Naive Bayes, with geospatial analysis using a 100×100 meter grid and the processing of georeferenced images. This solution facilitates the characterization and analysis of street vendor occupancy, the identification of patterns, the estimation of probabilities of vendor presence, and the creation of interactive visual representations. Although initial evaluations were performed using simulated data and labels, cross-validation confirms the correct execution of the workflow, thereby establishing a basis for the integration of actual data, the incorporation of deep learning techniques, and the applicability of the methodology to other urban environments. Accordingly, the objective is to foster an equilibrium between the socio-economic functions of informal trade and the structuring of public spaces. To achieve this objective, the following principal steps are considered: 1. Design and develop a prototype interface that consolidates and displays the collected information, highlighting areas with the highest occupation and their most relevant attributes. 2. Implement a data loading mechanism from multiple sources (images, official records), enabling their processing with computer vision algorithms and geospatial analysis. 3. Create result visualizations, incorporating charts and maps to highlight patterns and trends in the distribution of informal trade. 4. Integrate machine learning methods that facilitate the classification and segmentation of information, promoting evidence-based decision-making. 5. Conduct validation tests to ensure the accuracy and consistency of the algorithms, evaluating their efficiency in identifying and analyzing informal trade patterns. This multidisciplinary approach, a pioneering initiative in Chile, enhances understanding of a phenomenon affecting socio-economic development and urban planning. It provides a robust tool for the management of informal trade and the formulation of evidence-based public policy.
Informal trade within urban settings presents significant challenges to the management of public spaces and the formulation of decisions grounded in empirical evidence. In response to this issue in the context of downtown Santiago, this study develops a comprehensive platform integrating artificial intelligence methodologies, specifically automated vision via YOLO and probabilistic classification using Naive Bayes, with geospatial analysis using a 100×100 meter grid and the processing of georeferenced images. This solution facilitates the characterization and analysis of street vendor occupancy, the identification of patterns, the estimation of probabilities of vendor presence, and the creation of interactive visual representations. Although initial evaluations were performed using simulated data and labels, cross-validation confirms the correct execution of the workflow, thereby establishing a basis for the integration of actual data, the incorporation of deep learning techniques, and the applicability of the methodology to other urban environments. Accordingly, the objective is to foster an equilibrium between the socio-economic functions of informal trade and the structuring of public spaces. To achieve this objective, the following principal steps are considered: 1. Design and develop a prototype interface that consolidates and displays the collected information, highlighting areas with the highest occupation and their most relevant attributes. 2. Implement a data loading mechanism from multiple sources (images, official records), enabling their processing with computer vision algorithms and geospatial analysis. 3. Create result visualizations, incorporating charts and maps to highlight patterns and trends in the distribution of informal trade. 4. Integrate machine learning methods that facilitate the classification and segmentation of information, promoting evidence-based decision-making. 5. Conduct validation tests to ensure the accuracy and consistency of the algorithms, evaluating their efficiency in identifying and analyzing informal trade patterns. This multidisciplinary approach, a pioneering initiative in Chile, enhances understanding of a phenomenon affecting socio-economic development and urban planning. It provides a robust tool for the management of informal trade and the formulation of evidence-based public policy.
Description
Keywords
Comercio informal, Análisis geoespacial, Aprendizaje automático, SIG, Planificación urbana
