| Title |
Spatial time series prediction using bayesian network models / |
| Translation of Title |
Erdvinių laiko eilučių modeliavimas naudojant Bajeso tinklų metodus. |
| Authors |
Gertaitė, Justina |
| Full Text |
|
| Pages |
65 |
| Keywords [eng] |
Bayesian Network, spatial time series, Bayesian Network with residual correction mechanism, spatial Bayesian Network, new fuzzy Bayesian Network, DBSCAN, wild bootstrap. |
| Abstract [eng] |
This Master's thesis is based on book 'Enhanced Bayesian Network Models for Spatial Time Series Prediction'. Methods written in the mentioned book as Bayesian Network with Residual Correction Mechanism, Spatial Bayesian Network and New Fuzzy Bayesian Networks were defined. DBSCAN clustering algorithm was integrated into Spatial Bayesian Network to choose locations. Wild bootstrap was applied to improve predictions. Five different Bayesian Networks were compared on spatial time series data set, where precipitation and weather events of Unites States of America were predicted. The proposed models could help insurers or skiing resort owners to evaluate prices, insurance reserves or season length of skiing resorts. However, mentioned models should be applied on a different data set or longer time series as predictions of thesis are very weak. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2023 |
