/Bayesian-mixture-model-for-environmental-application

This project was realized for the Bayesian Statistics course, held at Politecnico di Milano, A.Y. 2022/2023.

Primary LanguageR

Bayesian Mixture Model For Environmetal Application

Project for the course of Bayesian Statistics for Mathematical Engineering, academic year 2022-2023.

Project supervisors: Raffaele Argiento, Lucia Paci, Sirio Legramanti.

Team members: P. Bogani, P. Botta, S. Caresana, R. Carrara, G. Corbo, L. Mainini

Description

The World Health Organization considers air pollution a major global environmental risk to human health. Only in the EU in 2020, a total of 238 000 premature deaths were linked to exposure to particulate matter. Our objective is to develop Bayesian-mixture-model-based clustering algorithms for environmental applications. Specifically, we focused our attention on PM10.

Markdown

In the files Case study.Rmd, Data exploration.Rmd and Simulations.Rmd is possible to find all the code to reproduce the analysis done and described in the report.

Documents

In the folder documents are located the report of the project, the final presentation exposed on 14 February 2023 and the three pdf files obtained from the Rmarkdown of the three main chapters.

Code

The folder code contains:

  • tseriesclust.R and tseriesclust_first.R, the scripts containing the 2 algortihms implemented
  • include_all.R, a file that includes all the libraries and codes
  • The folder results containing the RData of the results and a file .R with the same script of the corresponding Rmarkdown
  • The folder simulations containing the file to generate the synthetic data, a folder with the codes to check the distribution of the full conditionals and a file .R with the same script of the corresponding Rmarkdown
  • The folder data exploration containing two file .R with the same script of the corresponding Rmarkdown
  • The folder auxiliary_functions containing the auxiliary functions for the models.
  • The folder data containing the datasets containing the timeseries and the stations' information.

All the code should be run setting the main folder of the repository as working directory.

References

In the folder references are located all the references we consulted to develop the analysis.