My two collegues and I worked on this project for the course "Bayesian Learning Montecarlo Simulation". You can find the details in the file "Report.pdf".
Source: https://www.kaggle.com/code/lucasar/analysis-of-forest-fires-in-brazil-1998-2017
Name of the file: amazon.csv
Short description: This dataset reports the number of forest fires in Brazil divided by state. The series comprises the period of approximately 10 years (1998 to 2017).
Task: Fit a regression model with the number of fires as the response variable. You can start by considering data from only one state. The two basic covariates are the year and the month. You can try using an integer-valued distribution, or, if that fails, a normal model (scaling the data with some transformation if necessary). If you want to use integer-valued observations, round the data to obtain integer numbers for the number of fires (since some data are not integers). You can also try a more complex model including data from more states.