Bayesian Networks Assignment 2

This repo contains the code developed for the second assignment of the Bayesian Networks & Causal Inference course at the Radboud University (NWI-IMC012).

Brief Introduction

In recent years, forest fires have been occurring more frequently and more intensely. Not only does this result in increased environmental damage, but these forest fires also become harder to fight. Therefore, understanding the factors that influence such forest fires is important, and these insights might also be used to predict the severity of forest fires. In the first assignment we set out to create a model based on prior knowledge that best captures the information in the data. We used the model to predict the size of the burned area. In the current assignment the goal was to learn the model by applying two different algorithms, the SI-HITON-PC algorithm and the Tabu Search Algorithm.

Structure of this Repo

This repo contains the following folders:

• evaluation contains plots of the metrics used to evaluate the graphs generated by the algorithms.
• graphs contains plots of the generated graphs

This repo also contains the following important files:

• assignment2.R contains the fundamental code regarding the project: Building the Bayesian network, fitting it, performing predictions and computing path coefficients.
• data_exploration.R contains all the code that we used for exploring and preprocessing the dataset.
• forestfires.csv is the original dataset, as it is publicly available on the UCI Machine Learning Repository. This data is used in data_exploration.R.
• explored_forestfires.csv contains the preprocessed data, after executing data_exploration.R. This file is used in assignment2.R.

Collaboration

This project has been done in a team of three people:

• Avuerro
• ElFilosofoX
• JordAI