Institutional Repository of Vilnius University
Smulkiausio lygio populiacijos prognozavimas naudojant erdvinius duomenis /
Abstract
dc:descriptionThe aim of this bachelor’s thesis is to analyse and compare the ability of three different methods ‑ random forest algorithm, spatial regression and graph neural networks ‑ to predict a population using spatial data. To achieve this objective, a comprehensive analysis of the scientific literature was first carried out to identify the most relevant knowledge and gaps related to existing population estimation methods. Alternative methodologies and their applicability to population estimation are also analysed. Next, the data production process is described, using various spatial and census data for the territory of Lithuania. The techniques used to construct the dataset and to extract additional spatial features are presented. After testing the different methods, the results are analysed according to predefined criteria and the advantages and disadvantages of each method are examined. The results showed that spatial regression was the most effective in explaining the data, while graph neural networks showed the poorest results. It was also observed that although most of the predictions were stable, in areas with high population densities the methods performed poorly.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Blažauskas, Tomas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210645223&prefLang=en_US