University of Twente Student Theses
Preparing for wildfires avoidance by analyzing land cover maps
Geest, J.S.T. van der (2022) Preparing for wildfires avoidance by analyzing land cover maps.
PDF
1MB |
Abstract: | The aim of this project was to find a method that is best suitable to predict wildfire susceptibility in Cyprus. This method needed to take factors into account on which it could base its predictions. Therefore, the selection of these factors was also of interest for this project. Based on background research that was conducted at the beginning of this project a list of 12 factors was created. This list was turned into a dataset containing the information of each factor for different data points in Cyprus. The dataset was then used to train two models, logistic regression and random forest. In the end, the logistic regression model produced an overall accuracy of 70%, the AUC of the model was 0.69 meaning that it is decently able to make the distinction between the positive (wildfire occurred) class and negative (no wildfire occurred) class. However, preconditions were set of an accuracy above 80% which resulted in the model not satisfying as a method suitable to predict wildfire susceptibility in Cyprus. Random forest on the other hand had an overall accuracy of 88% with an AUC of 0.89 and therefore did meet the precondition of accuracy above 80%. |
Item Type: | Essay (Bachelor) |
Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
Subject: | 31 mathematics, 54 computer science, 74 (human) geography, cartography, town and country planning, demography |
Programme: | Creative Technology BSc (50447) |
Link to this item: | https://purl.utwente.nl/essays/92449 |
Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Repository Staff Only: item control page