dc.contributor.author |
Στεφανή, Ευαγγελία
|
el |
dc.contributor.author |
Stefani, Evangelia
|
en |
dc.date.accessioned |
2022-04-13T05:52:48Z |
|
dc.identifier.uri |
https://dspace.lib.ntua.gr/xmlui/handle/123456789/55081 |
|
dc.identifier.uri |
http://dx.doi.org/10.26240/heal.ntua.22779 |
|
dc.rights |
Default License |
|
dc.subject |
Πρόβλεψη Κυκλοφορίας |
el |
dc.subject |
Χωροχρονικες συσχετίσεις |
|
dc.subject |
|
|
dc.title |
Πρότυπα Πρόβλεψης Κυκλοφορίας με Χωρο-Χρονικές Συσχετίσεις |
el |
heal.type |
bachelorThesis |
|
heal.classification |
Κυκλοφοριακή Τεχνική |
el |
heal.dateAvailable |
2023-04-12T21:00:00Z |
|
heal.language |
el |
|
heal.access |
embargo |
|
heal.recordProvider |
ntua |
el |
heal.publicationDate |
2021-11-04 |
|
heal.abstract |
This thesis explores the applicability and accuracy of a Long-Short Term Model (LSTM), which is enhanced with spatio-temporal correlation information based on Information Theoretic metrics. The model is developed and trained using traffic and public transport data from central Athens. Results indicate that the LSTM model provided a higher forecasting accuracy for short-term predictions when nonlinear spatio-temporal correlations were taken into account. Also, more advanced hyper parameter optimization and feature engineering methods would further improve the machine learning model.
Keywords: |
en |
heal.advisorName |
Βλαχογιάννη, Ελένη |
el |
heal.committeeMemberName |
Γιαννής, Γεώργιος |
el |
heal.committeeMemberName |
Κεπαπτσόγλου, Κωνσταντίνος |
el |
heal.academicPublisher |
Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Πολιτικών Μηχανικών |
el |
heal.academicPublisherID |
ntua |
|
heal.fullTextAvailability |
false |
|