Biblio

Predicting daily air temperatures by Support Vector Machines Regression

TitlePredicting daily air temperatures by Support Vector Machines Regression
Publication TypeConference Paper
Year of Publication2014
AuthorsMarjanović M
Conference NameDailyMeteo.org/2014
Date Published06/2014
PublisherFaculty of Civil Engineering, University of Belgrade
Conference LocationBelgrade
Abstract

This paper demonstrates an attempt to apply Support Vector Machines regression (SVMr) on meteorological data. The objective was to predict daily air temperatures for continental part of Europe for 1.1.2011. Supplied training set was based on temperature records of 357 weather stations throughout Europe and additional attributes (extracted from appropriate grid maps), comprising of MODIS satellite image for 1.1.2011, elevation, coastline distance and surface insolation. The SVMr algorithm was then optimized by means of 10-fold Cross Validation and obtained parameters were used to learn a regression rule and expand it to the entire continent. The chosen data resolution of 0.036 arc degrees (approximately 4 km) was sufficient to give rough temperature estimation on the specified date. A separate validation set of another 200 weather stations throughout Europe was supplied to evaluate the modeling performance. Resulting RMSE of about ±2°C proved that there is a good potential of applying SVMr or similar Machine Learning based techniques for extrapolating measured temperature data.

Attachment: 

AttachmentSize
Presentation5.78 MB

Event Calendar

S M T W T F S
 
 
 
 
1
 
2
 
3
 
4
 
5
 
6
 
7
 
8
 
9
 
10
 
11
 
12
 
13
 
14
 
15
 
16
 
17
 
18
 
19
 
20
 
21
 
22
 
23
 
24
 
25
 
26
 
27
 
28
 
29
 
30
 
 

Meteo package

Recent comments

No comments available.