Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/107026
Title: Predicción de la calidad del aire de la ciudad de Medellín y su área metropolitana mediante el uso de redes neuronales recurrentes
Author: Maestre Sanmiguel, Oscar Jovanni
Director: Casas-Roma, Jordi  
Tutor: Parada Medina, Raúl  
Abstract: Recent years the levels of air pollution have increased in Medellin and its metropolitan area, the cause is largely due to the emission of agents harmful to human health. WHO in its air quality guidelines refers to four common pollutants: particulate matter (PM), ozone (O3), nitrogen dioxide (NO2) and sulfur dioxide (SO2). In the particular case of the city of Medellín and its metropolitan area, there has been an increase in acute respiratory diseases; studies carried out by various institutions between 2017 and 2018 indicated that the emission of PM2.5 particulate material amounted to 1,230 tons per year from mobile (motor vehicles) and fixed sources (factories), each one with a representation of 70% and 30% respectively. The mitigation measures adopted by the municipal administrations are reactive and unplanned, mainly focused on restricting vehicle traffic during some hours of the day. Due to the above, the objective of this master's final project is to predict the levels of PM2.5 particulate material 48 hours in advance using recurrent neural network models as RNN, GRU, LSTM and one hybrid model that combines LSTM and MLP. In the models training was performed with open data supplied by the Medellín Early Warning System and the Aburrá Valley ¿ SIATA, which a set of techniques and processes of data mining was applied. The results are compared based on the accuracy of the predictions of the generated models, and these are evaluated to determine if they are useful as a tool to help in decision-making, which allow the implementation of corrective measures by the responsible public administrations.
Keywords: recurrent neural networks
data mining
RNN
air quality
GRU
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 8-Jan-2020
Publication license: http://creativecommons.org/licenses/by-nc-sa/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
TFM_notebooks_datos.zipJupyter Notebooks7,9 MBUnknownView/Open
omaestresTFM0120memoria.pdfMemoria del TFM9,54 MBAdobe PDFThumbnail
View/Open
omaestresTFM0120presentación.pdfPresentación del TFM1,57 MBAdobe PDFThumbnail
View/Open