Comparación de algoritmos de Machine Learning para aplicaciones ambientales
Cargando...
Fecha
2022
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Nacional Arturo Jauretche
Resumen
El presente informe es un trabajo de investigación enfocado en las bases de Machine Learning y su aplicación para resolver dos ejercicios de predicción. El primero consiste en calcular cuánto cobrará una aseguradora basado en los datos personales del individuo; el segundo consiste en clasificar un celular en un rango de precio dependiendo de sus especificaciones técnicas. Los algoritmos utilizados para la resolución de la tarea pertenecen a tres tipos de modelos: de Regresión, de Clasificación y de Redes Neuronales Artificiales. En total se utilizaron un total de cinco tipos de algoritmos: Regresión lineal múltiple, K-NN, Árbol de decisión, Bosques aleatorios, y Redes Neuronales Artificiales. Además, se demostró cómo analizar y preprocesar los datos para aumentar significativamente el rendimiento de los algoritmos. El resultado de la investigación retornó en todos los modelos un porcentaje de éxito que oscila entre el 79% y 96%. El informe tiene como principal función actuar como guía e introducción para los estudiantes e interesados en adentrarse al mundo de Machine Learning, tanto de manera teórica como práctica utilizando casos del mundo real.
-
This report is a research work focused on the foundations of Machine Learning and its application to solve two prediction exercises. The first consists of calculating how much an insurer would charge based on the individual's personal data; the second consists of classifying a cellphone in a price range depending on its technical specifications. The algorithms used to solve the task belong to three types of models: Regression, Classification and Artificial Neural Networks. A total of five types of algorithms were used: Multiple Linear Regression, K-NN, Decision Tree, Random Forests, and Artificial Neural Networks. In addition, it was demonstrated how to analyze and preprocess the data to significantly increase the performance of the algorithms. The result of the investigation returned in all the models a percentage of success that oscillates between 79% and 96%. The report is helpful as a guide and introduction for students and those interested in entering the world of Machine Learning, both theoretically and practically using real-world cases.
This report is a research work focused on the foundations of Machine Learning and its application to solve two prediction exercises. The first consists of calculating how much an insurer would charge based on the individual's personal data; the second consists of classifying a cellphone in a price range depending on its technical specifications. The algorithms used to solve the task belong to three types of models: Regression, Classification and Artificial Neural Networks. A total of five types of algorithms were used: Multiple Linear Regression, K-NN, Decision Tree, Random Forests, and Artificial Neural Networks. In addition, it was demonstrated how to analyze and preprocess the data to significantly increase the performance of the algorithms. The result of the investigation returned in all the models a percentage of success that oscillates between 79% and 96%. The report is helpful as a guide and introduction for students and those interested in entering the world of Machine Learning, both theoretically and practically using real-world cases.