Desarrollo de un sistema recomendador de licitaciones basado en tecnica de aprendizaje automatico
Fecha
2022
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad de Guayaquil. Facultad de Ciencias Matemáticas y Físicas. Carrera de Ingeniería en Networking y Telecomunicaciones
Resumen
El presente proyecto tiene como objetivo desarrollar e implementar un sistema web que recomiende a una empresa participar o no en una licitación basándose en técnica de aprendizaje automático. Para ello se definió a que tipo de empresas será dirigido este proyecto que serían las PYMES (pequeñas y medianas empresas). El estudio está basado en las licitaciones de España ya que el sitio de contrataciones públicas de dicho país permite la descarga de datos abiertos de forma rápida y fácil.
La ventaja de los resultados obtenidos del estudio es que puede ser aplicable en otros países porque por lo general, se mantiene el mismo proceso de contratación pública.
Mediante esta base de datos se definió las variables de entrada y salida, una vez realizado ese filtro se estudió varios algoritmos de Machine Learning como: Maquina de soporte vector, Regresión logística, Árbol de decisiones, Bosque aleatorio, K vecinos, Naive Bayes. En la cual se optó por tres algoritmos para comparar resultados que serían desde el mejor resultado de probabilidad hasta el menor, esos serian: Bosque aleatorio, en medio árbol de decisiones y con porcentaje bajo K- vecinos. Una vez realizado el estudio de los modelos de aprendizaje se procedió a crear un sistema web realizado en la plataforma Figma en donde se ingresará los datos de entrada y nos mostrará el resultado si adjudico o no en la licitación.
The present project aims to develop and implement a web system that recommends a company or not in a tender based on automatic learning technique. For this, this type of companies was defined to this project that SMEs (small and medium enterprises) will be directed. The study is based on the tenders of Spain since the public hiring site of that country allows the discharge of open data from Fast and easy way. The advantage of the results obtained from the study is that it can be applicable in other countries because in general, the same public procurement process is maintained. Through this database, the input and output variables were defined, once this filter was carried out, several Machine Learning algorithms such as: Vector Support Machine, Logistics Regression, Decision Tree, Random Forest, K Neighbors, Naive Bayes were studied. In which three algorithms were chosen to compare results that would be from the best probability result to the child, those would be: random forest, in the midst of decisions and with low-neighbors. Once the study of the learning models was carried out, a web system carried out on the Figma platform was created where the input data will be entered and the result will show us whether or not to award the tender.
The present project aims to develop and implement a web system that recommends a company or not in a tender based on automatic learning technique. For this, this type of companies was defined to this project that SMEs (small and medium enterprises) will be directed. The study is based on the tenders of Spain since the public hiring site of that country allows the discharge of open data from Fast and easy way. The advantage of the results obtained from the study is that it can be applicable in other countries because in general, the same public procurement process is maintained. Through this database, the input and output variables were defined, once this filter was carried out, several Machine Learning algorithms such as: Vector Support Machine, Logistics Regression, Decision Tree, Random Forest, K Neighbors, Naive Bayes were studied. In which three algorithms were chosen to compare results that would be from the best probability result to the child, those would be: random forest, in the midst of decisions and with low-neighbors. Once the study of the learning models was carried out, a web system carried out on the Figma platform was created where the input data will be entered and the result will show us whether or not to award the tender.
Descripción
PDF
Palabras clave
Licitaciones, Machine Learning, Pymes, Algoritmo