Implementation of a Predictive Modeling for Pollution Levels in Barcelona
Visualitza/Obre
tfg-report.pdf (8,642Mb) (Accés restringit)
tfg-appendices.pdf (8,156Mb) (Accés restringit)
Estadístiques de LA Referencia / Recolecta
Inclou dades d'ús des de 2022
Cita com:
hdl:2117/330242
Tipus de documentTreball Final de Grau
Data2020-07-21
Condicions d'accésAccés restringit per decisió de l'autor
Tots els drets reservats. Aquesta obra està protegida pels drets de propietat intel·lectual i
industrial corresponents. Sense perjudici de les exempcions legals existents, queda prohibida la seva
reproducció, distribució, comunicació pública o transformació sense l'autorització del titular dels drets
Abstract
This project consists in a computer application developed to predict pollution levels for subsequent days in Barcelona by applying Artificial Intelligence algorithms. These predictions may be used by public administrations in order to avoid exceeding the limited values required.
Firstly, data will be collected from different data sources. It has been proved that pollution values depend on meteorology and traffic occupancy. Hence, the application will be using pollution data, meteorology measurements and road traffic occupancy in order to compute the predicted values. Historical data from 2012 will be used in order to analyze these dependencies and train the model. The pollutants analyzed will be CO, NO2, SO2, O3 and PM10.
Secondly, some Machine Learning regression models will be compared to get the one computing the best predictions. Mean absolute error will be used to compare all algorithms. It will be concluded that Random Forest is the model producing the best performance. The application will predict values for four subsequent days.
The model will be predicting pollution values located in concrete points from Barcelona, which correspond to the positions where pollution stations are. The aim of this project is to get a complete distribution of pollution levels around the whole city. Inverse Distance Weighting will be applied to interpolate the values and obtain an accurate pollution distribution.
Finally, all values obtained will be presented using a Dashboard. It will consist of a heat map so that areas with high pollution levels will be filled with an intense color. Moreover, the user will be able to select which pollutant wants to be represented in the Dashboard and the date of the predictions shown.
This application can be used by public administrations in order to take measures when high pollution levels are predicted. Intensifying teleworking or reducing private transport by increasing public transport are just two examples of possible measures that could be taken. Moreover, people could also use this information in order to identify green routes for running, walking or biking, among other utilities.
TitulacióGRAU EN ENGINYERIA EN TECNOLOGIES INDUSTRIALS (Pla 2010)
Fitxers | Descripció | Mida | Format | Visualitza |
---|---|---|---|---|
tfg-report.pdf | 8,642Mb | Accés restringit | ||
tfg-appendices.pdf | 8,156Mb | Accés restringit |