Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
A Comparative Analysis of RNN and SVM: Electricity Price Forecasting in Energy Management Systems
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2019 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
En jämförande analys av RNN och SVM : Prognos för elpriser i energiledningssystem (Swedish)
Abstract [en]

A trend in increasing electricity consumption and technological innovation has resulted in automated energy management systems. Forecasting movement of the electricity price with machine learning plays a role in the sustainability of these systems.

The aim of the report is to compare two machine learning methods, namely Recurrent Neural Network (RNN) with LSTM, and Support Vector Machine (SVM). The metric to evaluate is percentage in prediction accuracy, additionally statistical analysis is applied for further evaluation.

The models are built and optimized on a single historic dataset from an Australian electricity market where the major influencing attributes are price, demand and time. The training and test set are split 80/20 whereas the training is done in 10 folds for cross validation.

Results of the experiment show that the SVM-model had a slightly higher accuracy and a lower standard error of the mean. Differences were seen in sensitivity and specificity when applied to a confusion matrix.

The conclusion made was that in this specific case, SVM outperformed RNN in prediction accuracy, however, there is room for improvement of both implementations of these methods which could lead to a different result. In regard to specificity and sensitivity the choice of an SVM or RNN would be highly dependent on the implementation of real-world application.

Abstract [sv]

En trend i ökad elförbrukning och teknisk innovation har resulterat i automatiserade energiledningssystem. Prognos i förändringen av elpriser med maskininlärning spelar en roll för hållbarheten av dessa system.

Syftet med denna rapport är att jämföra de två maskininlärningsmetoderna, Reccurent Neural Network (RNN) med LSTM och Support Vector Machine (SVM). De värden som utvärderas är procentenheter i förutsägbarhetsnoggrannhet där statistisk analys tillämpas för ytterligare utvärdering.

Modellerna är byggda på historisk data från en australisk elmarknad där de väsentligaste egenskaperna är pris, efterfrågan och tid. Tränings- och testuppsättningen delas 80/20 och träningen görs med 10-delad korsvalidering.

Resultaten från analysen visar att SVM-metoden hade en något högre noggrannhet och lägre standardfel. Från en diagnostisk beslutsmatris beräknades sensitivitet och specificitet, i dessa värden upptäcktes skillnader.

Slutsatsen i vårt fall var att SVM är mer noggrann än RNN. Vi anser att utrymme för förbättring av båda modellerna finns, vilket kan leda till ett annat resultat. När det gäller sensitivitet och specificitet skulle valet av RNN eller SVM vara starkt beroende på tillämpningen av en verklig applikation.

Place, publisher, year, edition, pages
2019. , p. 29
Series
TRITA-EECS-EX ; 2019:353
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-259745OAI: oai:DiVA.org:kth-259745DiVA, id: diva2:1353342
Supervisors
Examiners
Available from: 2019-09-24 Created: 2019-09-23 Last updated: 2019-09-24Bibliographically approved

Open Access in DiVA

fulltext(2232 kB)8 downloads
File information
File name FULLTEXT01.pdfFile size 2232 kBChecksum SHA-512
9d0ae146272861a1a6928289c812e25c6d931d21841a2a433d4b2e11c22b843a5cbac40b86b6e8a0c0d192a28606004f5f19833fe247d719f116bf87f05dab78
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 8 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 111 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf