Data augmentation for time series regression: Applying transformations, autoencoders and adversarial networks to electricity price forecasting
A model’s expected generalisation error is inversely proportional to its training set size. This relationship can pose a problem when modelling multivariate time series, because structural breaks, low sampling rates, and high data gathering costs can severely restrict training set sizes, increasing...
Uloženo v:
| Vydáno v: | Applied energy Ročník 304; s. 117695 |
|---|---|
| Hlavní autoři: | , , , |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
Elsevier Ltd
15.12.2021
|
| Témata: | |
| ISSN: | 0306-2619, 1872-9118 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Abstract | A model’s expected generalisation error is inversely proportional to its training set size. This relationship can pose a problem when modelling multivariate time series, because structural breaks, low sampling rates, and high data gathering costs can severely restrict training set sizes, increasing a model’s expected generalisation error by spurring regression model overfitting. Artificially expanding the training set size, using data augmentation methods, can, however, counteract the restrictions imposed by small sample sizes: increasing a model’s robustness to overfitting and boosting out-of-sample prediction accuracies. While existing time series augmentation methods have predominantly utilised feature space transformations to artificially expand training set sizes and boost prediction accuracies, we propose using autoencoders (AEs), variational autoencoders (VAEs) and Wasserstein generative adversarial networks with a gradient penalty (WGAN-GPs) for time series augmentation. To evaluate our proposed augmentors, as a case study we forecast Belgian and Dutch day-ahead electricity market prices using both autoregressive models and artificial neural networks. Overall, our results demonstrate that AEs, VAEs, and WGAN-GPs can significantly boost regression accuracies; on average decreasing benchmark model mean absolute errors by 2.23%, 2.73% and 2.97% respectively. Moreover, our results demonstrate that combining AE, VAE, and WGAN-GP generated time series can further boost regression accuracies; on average decreasing benchmark errors by 3.44%. As our proposed augmentors outperform existing augmentation methods, we strongly believe that both practitioners and researchers aiming to generate time series or reduce time series regression errors will find utility in our study.
•Novel time series augmentation methods, using generative models, are developed.•The viability of augmenting multivariate time series with exogenous inputs is shown.•Electricity price forecast accuracies are statistically significantly improved.•Generative augmentors are found to outperform feature space augmentors.•Combining data from multiple augmentors is found to yield further improvements. |
|---|---|
| AbstractList | A model’s expected generalisation error is inversely proportional to its training set size. This relationship can pose a problem when modelling multivariate time series, because structural breaks, low sampling rates, and high data gathering costs can severely restrict training set sizes, increasing a model’s expected generalisation error by spurring regression model overfitting. Artificially expanding the training set size, using data augmentation methods, can, however, counteract the restrictions imposed by small sample sizes: increasing a model’s robustness to overfitting and boosting out-of-sample prediction accuracies. While existing time series augmentation methods have predominantly utilised feature space transformations to artificially expand training set sizes and boost prediction accuracies, we propose using autoencoders (AEs), variational autoencoders (VAEs) and Wasserstein generative adversarial networks with a gradient penalty (WGAN-GPs) for time series augmentation. To evaluate our proposed augmentors, as a case study we forecast Belgian and Dutch day-ahead electricity market prices using both autoregressive models and artificial neural networks. Overall, our results demonstrate that AEs, VAEs, and WGAN-GPs can significantly boost regression accuracies; on average decreasing benchmark model mean absolute errors by 2.23%, 2.73% and 2.97% respectively. Moreover, our results demonstrate that combining AE, VAE, and WGAN-GP generated time series can further boost regression accuracies; on average decreasing benchmark errors by 3.44%. As our proposed augmentors outperform existing augmentation methods, we strongly believe that both practitioners and researchers aiming to generate time series or reduce time series regression errors will find utility in our study. A model’s expected generalisation error is inversely proportional to its training set size. This relationship can pose a problem when modelling multivariate time series, because structural breaks, low sampling rates, and high data gathering costs can severely restrict training set sizes, increasing a model’s expected generalisation error by spurring regression model overfitting. Artificially expanding the training set size, using data augmentation methods, can, however, counteract the restrictions imposed by small sample sizes: increasing a model’s robustness to overfitting and boosting out-of-sample prediction accuracies. While existing time series augmentation methods have predominantly utilised feature space transformations to artificially expand training set sizes and boost prediction accuracies, we propose using autoencoders (AEs), variational autoencoders (VAEs) and Wasserstein generative adversarial networks with a gradient penalty (WGAN-GPs) for time series augmentation. To evaluate our proposed augmentors, as a case study we forecast Belgian and Dutch day-ahead electricity market prices using both autoregressive models and artificial neural networks. Overall, our results demonstrate that AEs, VAEs, and WGAN-GPs can significantly boost regression accuracies; on average decreasing benchmark model mean absolute errors by 2.23%, 2.73% and 2.97% respectively. Moreover, our results demonstrate that combining AE, VAE, and WGAN-GP generated time series can further boost regression accuracies; on average decreasing benchmark errors by 3.44%. As our proposed augmentors outperform existing augmentation methods, we strongly believe that both practitioners and researchers aiming to generate time series or reduce time series regression errors will find utility in our study. •Novel time series augmentation methods, using generative models, are developed.•The viability of augmenting multivariate time series with exogenous inputs is shown.•Electricity price forecast accuracies are statistically significantly improved.•Generative augmentors are found to outperform feature space augmentors.•Combining data from multiple augmentors is found to yield further improvements. |
| ArticleNumber | 117695 |
| Author | Demir, Sumeyra Paterakis, Nikolaos G. Mincev, Krystof Kok, Koen |
| Author_xml | – sequence: 1 givenname: Sumeyra orcidid: 0000-0002-4907-8058 surname: Demir fullname: Demir, Sumeyra email: s.demir@tue.nl organization: Department of Electrical Engineering, Eindhoven University of Technology, The Netherlands – sequence: 2 givenname: Krystof orcidid: 0000-0002-1686-8707 surname: Mincev fullname: Mincev, Krystof organization: School of Computer Science, University of St Andrews, United Kingdom – sequence: 3 givenname: Koen orcidid: 0000-0002-7979-213X surname: Kok fullname: Kok, Koen organization: Department of Electrical Engineering, Eindhoven University of Technology, The Netherlands – sequence: 4 givenname: Nikolaos G. orcidid: 0000-0002-3395-8253 surname: Paterakis fullname: Paterakis, Nikolaos G. organization: Department of Electrical Engineering, Eindhoven University of Technology, The Netherlands |
| BookMark | eNqFUU1vGyEQRZErxUn6FyKOOWRdYL1gqh5iuV-RIuWSnhFmZy2cXdgy2JH_RH9zcZxeeonmMCPx3hvmvQsyCTEAIdeczTjj8tN2ZkcIkDaHmWCCzzhXUjdnZMoXSlSa88WETFnNZCUk1-fkAnHLWEEKNiV_vtpsqd1tBgjZZh8D7WKi2Q9AEZIHpAk2CRDL02e6HMf-4MOG5mQDFuTwysHbIpEjBBdbSEhtaKlt92W0ydueBsgvMT0jzZFCDy4n73w-0LF0OC4EZzEX3SvyobM9wse3fkl-ff_2tPpZPTz-uF8tHyo3b3SuOtANAAhQUjEmWVc3MO-UamotuBOycV3nSrG1XrRrWAvt5k7N-RoaDla6-pLcnHTHFH_vALMZPDroexsg7tAIWctGCa1VgX45QV2KiAk6U77-enXxwPeGM3OMwWzNvxjMMQZziqHQ5X_0cvRg0-F94t2JCMWHvYdk0PniMLS-2JVNG_17En8BbVuuSw |
| CitedBy_id | crossref_primary_10_1016_j_gsf_2023_101670 crossref_primary_10_1016_j_rser_2025_115817 crossref_primary_10_3390_en16196767 crossref_primary_10_2478_amns_2025_0845 crossref_primary_10_1109_TSG_2024_3355805 crossref_primary_10_2166_wcc_2025_849 crossref_primary_10_1109_JSEN_2024_3483108 crossref_primary_10_1016_j_asoc_2025_113112 crossref_primary_10_3390_bioengineering9100529 crossref_primary_10_1109_ACCESS_2023_3258544 crossref_primary_10_26599_BDMA_2023_9020028 crossref_primary_10_1109_TPWRS_2024_3364166 crossref_primary_10_1007_s12273_024_1143_4 crossref_primary_10_1016_j_apenergy_2024_123233 crossref_primary_10_1016_j_eneco_2022_106471 crossref_primary_10_1371_journal_pone_0316548 crossref_primary_10_3390_bioengineering10040405 crossref_primary_10_3390_w16121731 crossref_primary_10_1016_j_renene_2023_119374 crossref_primary_10_1016_j_apenergy_2022_118752 crossref_primary_10_1007_s00521_023_08459_3 crossref_primary_10_1109_TPWRS_2023_3301442 crossref_primary_10_1007_s10479_022_04857_3 crossref_primary_10_1016_j_ijepes_2024_110190 crossref_primary_10_1111_mice_13413 crossref_primary_10_1016_j_prime_2025_101024 crossref_primary_10_3390_bioengineering11080803 crossref_primary_10_1016_j_buildenv_2023_110149 crossref_primary_10_3390_en17061374 crossref_primary_10_1016_j_energy_2022_124212 crossref_primary_10_3390_make6020049 crossref_primary_10_1109_ACCESS_2025_3528450 crossref_primary_10_1016_j_ins_2025_122655 crossref_primary_10_1016_j_compag_2023_107646 crossref_primary_10_1016_j_oceaneng_2023_116137 crossref_primary_10_1109_ACCESS_2023_3253429 crossref_primary_10_1016_j_applthermaleng_2023_121545 crossref_primary_10_1109_TIM_2024_3378269 crossref_primary_10_1002_pol_20230649 crossref_primary_10_3390_app122412793 crossref_primary_10_1016_j_buildenv_2023_110407 crossref_primary_10_1016_j_asoc_2025_113012 crossref_primary_10_1016_j_apenergy_2024_123058 crossref_primary_10_1016_j_energy_2023_128589 crossref_primary_10_1007_s10639_023_12334_y crossref_primary_10_1016_j_enbuild_2024_114027 crossref_primary_10_3390_computers11050058 crossref_primary_10_1007_s13369_024_08794_0 crossref_primary_10_1109_ACCESS_2025_3536287 crossref_primary_10_3390_s23042312 crossref_primary_10_1109_JSEN_2024_3521896 crossref_primary_10_1016_j_diamond_2025_112352 crossref_primary_10_3390_en18030742 crossref_primary_10_3389_fenrg_2022_952420 crossref_primary_10_1016_j_jpowsour_2023_233286 crossref_primary_10_1016_j_segan_2025_101865 crossref_primary_10_1016_j_segan_2023_101023 crossref_primary_10_3390_ma17051142 crossref_primary_10_1016_j_apenergy_2023_120711 crossref_primary_10_3390_coatings14030288 crossref_primary_10_1007_s11431_024_2917_8 crossref_primary_10_1016_j_energy_2022_124558 crossref_primary_10_1109_TIM_2025_3578691 |
| Cites_doi | 10.3390/app10010255 10.1016/j.epsr.2021.107416 10.1016/j.apenergy.2018.02.069 10.1109/ACCESS.2020.3048519 10.2307/3001968 10.3390/en11051255 10.1016/j.apenergy.2019.114087 10.1007/978-0-387-21606-5 10.1109/TIP.2018.2836316 10.1214/aoms/1177730491 10.1023/A:1010933404324 10.1016/j.ijforecast.2014.08.008 10.3390/en11082039 10.1016/j.apenergy.2016.12.130 10.3390/en9080621 10.1016/j.ijforecast.2015.07.002 10.1016/j.apenergy.2017.11.098 10.1016/j.apenergy.2021.116983 10.1080/12460125.2015.994290 10.1016/j.neucom.2018.09.013 10.1109/CVPR.2017.632 |
| ContentType | Journal Article |
| Copyright | 2021 The Authors |
| Copyright_xml | – notice: 2021 The Authors |
| DBID | 6I. AAFTH AAYXX CITATION 7S9 L.6 |
| DOI | 10.1016/j.apenergy.2021.117695 |
| DatabaseName | ScienceDirect Open Access Titles Elsevier:ScienceDirect:Open Access CrossRef AGRICOLA AGRICOLA - Academic |
| DatabaseTitle | CrossRef AGRICOLA AGRICOLA - Academic |
| DatabaseTitleList | AGRICOLA |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering Environmental Sciences |
| EISSN | 1872-9118 |
| ExternalDocumentID | 10_1016_j_apenergy_2021_117695 S0306261921010527 |
| GrantInformation_xml | – fundername: Scholt Energy |
| GroupedDBID | --K --M .~1 0R~ 1B1 1~. 1~5 23M 4.4 457 4G. 5GY 5VS 6I. 7-5 71M 8P~ 9JN AABNK AACTN AAEDT AAEDW AAFTH AAHCO AAIAV AAIKJ AAKOC AALRI AAOAW AAQFI AARJD AAXUO ABJNI ABMAC ABYKQ ACDAQ ACGFS ACRLP ADBBV ADEZE ADTZH AEBSH AECPX AEKER AENEX AFKWA AFTJW AGHFR AGUBO AGYEJ AHHHB AHIDL AHJVU AIEXJ AIKHN AITUG AJOXV ALMA_UNASSIGNED_HOLDINGS AMFUW AMRAJ AXJTR BELTK BJAXD BKOJK BLXMC CS3 EBS EFJIC EFLBG EO8 EO9 EP2 EP3 FDB FIRID FNPLU FYGXN G-Q GBLVA IHE J1W JARJE JJJVA KOM LY6 M41 MO0 N9A O-L O9- OAUVE OZT P-8 P-9 P2P PC. Q38 ROL RPZ SDF SDG SES SPC SPCBC SSR SST SSZ T5K TN5 ~02 ~G- 9DU AAHBH AAQXK AATTM AAXKI AAYWO AAYXX ABEFU ABFNM ABWVN ABXDB ACLOT ACNNM ACRPL ACVFH ADCNI ADMUD ADNMO AEIPS AEUPX AFJKZ AFPUW AGQPQ AIGII AIIUN AKBMS AKRWK AKYEP ANKPU APXCP ASPBG AVWKF AZFZN CITATION EFKBS EJD FEDTE FGOYB G-2 HVGLF HZ~ R2- SAC SEW WUQ ZY4 ~HD 7S9 L.6 |
| ID | FETCH-LOGICAL-c459t-fe95eee2e7670060f35e4f7753921c265cffcfcf0b98dbeb29c4c741be51ea6c3 |
| ISICitedReferencesCount | 76 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000701874500012&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 0306-2619 |
| IngestDate | Sun Sep 28 12:29:42 EDT 2025 Sat Nov 29 07:17:55 EST 2025 Tue Nov 18 22:35:12 EST 2025 Fri Feb 23 02:40:59 EST 2024 |
| IsDoiOpenAccess | true |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Keywords | Augmentation Adversarial network Electricity price forecasting Regression Multivariate time series Autoencoder |
| Language | English |
| License | This is an open access article under the CC BY license. |
| LinkModel | OpenURL |
| MergedId | FETCHMERGED-LOGICAL-c459t-fe95eee2e7670060f35e4f7753921c265cffcfcf0b98dbeb29c4c741be51ea6c3 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| ORCID | 0000-0002-4907-8058 0000-0002-3395-8253 0000-0002-7979-213X 0000-0002-1686-8707 |
| OpenAccessLink | https://dx.doi.org/10.1016/j.apenergy.2021.117695 |
| PQID | 2636572997 |
| PQPubID | 24069 |
| ParticipantIDs | proquest_miscellaneous_2636572997 crossref_citationtrail_10_1016_j_apenergy_2021_117695 crossref_primary_10_1016_j_apenergy_2021_117695 elsevier_sciencedirect_doi_10_1016_j_apenergy_2021_117695 |
| PublicationCentury | 2000 |
| PublicationDate | 2021-12-15 |
| PublicationDateYYYYMMDD | 2021-12-15 |
| PublicationDate_xml | – month: 12 year: 2021 text: 2021-12-15 day: 15 |
| PublicationDecade | 2020 |
| PublicationTitle | Applied energy |
| PublicationYear | 2021 |
| Publisher | Elsevier Ltd |
| Publisher_xml | – name: Elsevier Ltd |
| References | Arjovsky, Bottou (b33) 2017 Sun, Shrivastava, Singh, Gupta (b2) 2017 Mann, Whitney (b51) 1947 Gulrajani, Ahmed, Arjovsky, Dumoulin, Courville (b35) 2017 Vincent, Larochelle, Bengio, Manzagol (b28) 2008 Breiman (b42) 2001; 45 Kingma, Welling (b30) 2013 Goodfellow, Pouget-Abadie, Mirza, Xu, Warde-Farley, Ozair (b32) 2014; vol. 2 Smyl S, Kuber K. Data preprocessing and augmentation for multiple short time series forecasting with recurrent neural networks. In: 36th International Symposium on Forecasting. 2016. Cherkassky, Mulier (b39) 1998 (b36) 2020 Petzka, Fischer, Lukovnicov (b48) 2017 Goodfellow, Pouget-Abadie, Mirza, Xu, Warde-Farley, Ozair (b12) 2014; vol. 27 Weron (b16) 2014; 30 Yang, Ce, Lian (b24) 2017; 190 Uniejewski, Weron (b20) 2018; 11 Ugurlu, Oksuz, Tas (b21) 2018; 11 Bergmeir, Hyndman, Benítez (b9) 2016; 32 Uniejewski, Nowotarski, Weron (b19) 2016; 9 Demir, Mincev, Kok, Paterakis (b14) 2019; 10 Creswell, Arulkumaran, Bharath (b38) 2017 Hastie, Tibshirani, Friedman (b40) 2001 Lago, Ridder, Vrancx, Schutter (b17) 2018; 211 Goodfellow, Bengio, Courville (b27) 2016 Gunduz, Ugurlu, Oksuz (b22) 2020 Hinton, Zemel (b11) 1994; vol. 6 Isola P, Zhu J-Y, Zhou T, Efros AA. Image-to-image translation with conditional adversarial networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2017, p. 1125–34. Elattar, Elsayed, Farrag (b26) 2021; 9 Wang, Xu, Wang, Tao (b46) 2018; 27 Jorge, Vieco, Paredes, Sánchez, Benedí (b6) 2018 Jolliffe (b49) 2011 Luo, Lu (b7) 2018 Wilcoxon (b50) 1945; 1 Lago, Ridder, Schutter (b18) 2018; 221 Um, Pfister, Pichler, Endo, Lang, Hirche (b4) 2017 DeVries, Taylor (b5) 2017 Guennec AL, Malinowski S, Tavenard R. Data augmentation for time series classification using convolutional neural networks. In: ECML/PKDD workshop on advanced analytics and learning on temporal data. Riva Del Garda, Italy; 2016. Zhao, Mathieu, LeCun (b47) 2016 Abu-Mostafa, Magdon-Ismail, Lin (b1) 2012 Higgins, Matthey, Pal, Burgess, Glorot, Botvinick (b44) 2017 Hu, Yang, Salakhutdinov, Xing (b31) 2017 Lago, Marcjasz, De Schutter, Weron (b15) 2021; 293 Thornton, Hutter, Hoos, Leyton-Brown (b43) 2013 Frid-Adar, Diamant, Klang, Amitai, Goldberger, Greenspan (b13) 2018; 321 Tran, Pham, Carneiro, Palmer, Reid (b8) 2017 Ludwig, Feuerriegel, Neumann (b41) 2015; 24 Zhang, Tan, Wei (b25) 2020; 258 Shi, Wang, Chen, Ma (b23) 2021; 199 Arjovsky, Chintala, Bottou (b34) 2017 (b37) 2020 Bengio Y, Mesnil G, Dauphin Y, Rifai S. Better mixing via deep representations. In: International conference on machine learning. 2013, p. 552–60. 10.1016/j.apenergy.2021.117695_b10 Creswell (10.1016/j.apenergy.2021.117695_b38) 2017 Demir (10.1016/j.apenergy.2021.117695_b14) 2019; 10 Gunduz (10.1016/j.apenergy.2021.117695_b22) 2020 Thornton (10.1016/j.apenergy.2021.117695_b43) 2013 Mann (10.1016/j.apenergy.2021.117695_b51) 1947 Lago (10.1016/j.apenergy.2021.117695_b15) 2021; 293 Bergmeir (10.1016/j.apenergy.2021.117695_b9) 2016; 32 Weron (10.1016/j.apenergy.2021.117695_b16) 2014; 30 Uniejewski (10.1016/j.apenergy.2021.117695_b20) 2018; 11 Lago (10.1016/j.apenergy.2021.117695_b17) 2018; 211 Cherkassky (10.1016/j.apenergy.2021.117695_b39) 1998 Arjovsky (10.1016/j.apenergy.2021.117695_b34) 2017 Wang (10.1016/j.apenergy.2021.117695_b46) 2018; 27 Goodfellow (10.1016/j.apenergy.2021.117695_b27) 2016 Zhao (10.1016/j.apenergy.2021.117695_b47) 2016 Goodfellow (10.1016/j.apenergy.2021.117695_b12) 2014; vol. 27 Tran (10.1016/j.apenergy.2021.117695_b8) 2017 Um (10.1016/j.apenergy.2021.117695_b4) 2017 Hu (10.1016/j.apenergy.2021.117695_b31) 2017 10.1016/j.apenergy.2021.117695_b29 Vincent (10.1016/j.apenergy.2021.117695_b28) 2008 Petzka (10.1016/j.apenergy.2021.117695_b48) 2017 Zhang (10.1016/j.apenergy.2021.117695_b25) 2020; 258 Hastie (10.1016/j.apenergy.2021.117695_b40) 2001 (10.1016/j.apenergy.2021.117695_b36) 2020 Abu-Mostafa (10.1016/j.apenergy.2021.117695_b1) 2012 Gulrajani (10.1016/j.apenergy.2021.117695_b35) 2017 Frid-Adar (10.1016/j.apenergy.2021.117695_b13) 2018; 321 Hinton (10.1016/j.apenergy.2021.117695_b11) 1994; vol. 6 Shi (10.1016/j.apenergy.2021.117695_b23) 2021; 199 Sun (10.1016/j.apenergy.2021.117695_b2) 2017 Arjovsky (10.1016/j.apenergy.2021.117695_b33) 2017 Kingma (10.1016/j.apenergy.2021.117695_b30) 2013 Goodfellow (10.1016/j.apenergy.2021.117695_b32) 2014; vol. 2 Lago (10.1016/j.apenergy.2021.117695_b18) 2018; 221 Uniejewski (10.1016/j.apenergy.2021.117695_b19) 2016; 9 Breiman (10.1016/j.apenergy.2021.117695_b42) 2001; 45 Jolliffe (10.1016/j.apenergy.2021.117695_b49) 2011 10.1016/j.apenergy.2021.117695_b3 DeVries (10.1016/j.apenergy.2021.117695_b5) 2017 Jorge (10.1016/j.apenergy.2021.117695_b6) 2018 Higgins (10.1016/j.apenergy.2021.117695_b44) 2017 Wilcoxon (10.1016/j.apenergy.2021.117695_b50) 1945; 1 (10.1016/j.apenergy.2021.117695_b37) 2020 Elattar (10.1016/j.apenergy.2021.117695_b26) 2021; 9 Luo (10.1016/j.apenergy.2021.117695_b7) 2018 Ugurlu (10.1016/j.apenergy.2021.117695_b21) 2018; 11 Ludwig (10.1016/j.apenergy.2021.117695_b41) 2015; 24 Yang (10.1016/j.apenergy.2021.117695_b24) 2017; 190 10.1016/j.apenergy.2021.117695_b45 |
| References_xml | – volume: 10 year: 2019 ident: b14 article-title: Introducing technical indicators to electricity price forecasting: A feature engineering study for linear, ensemble, and deep machine learning models publication-title: Appl Sci – start-page: 2535 year: 2018 end-page: 2538 ident: b7 article-title: Eeg data augmentation for emotion recognition using a conditional wasserstein gan publication-title: 2018 40th annual international conference of the IEEE Engineering in Medicine and Biology Society (EMBC) – year: 2017 ident: b5 article-title: Dataset augmentation in feature space – volume: 190 start-page: 291 year: 2017 end-page: 305 ident: b24 article-title: Electricity price forecasting by a hybrid model, combining wavelet transform, ARMA and kernel-based extreme learning machine methods publication-title: Appl Energy – volume: 321 start-page: 321 year: 2018 end-page: 331 ident: b13 article-title: Gan-based synthetic medical image augmentation for increased cnn performance in liver lesion classification publication-title: Neurocomputing – volume: 30 start-page: 1030 year: 2014 end-page: 1081 ident: b16 article-title: Electricity price forecasting: A review of the state-of-the-art with a look into the future publication-title: Int J Forecast – year: 2016 ident: b27 article-title: Deep learning – year: 1998 ident: b39 article-title: Learning from data: concepts, theory, and methods – year: 2017 ident: b33 article-title: Towards principled methods for training generative adversarial networks – start-page: 5767 year: 2017 end-page: 5777 ident: b35 article-title: Improved training of wasserstein gans publication-title: Advances in neural information processing systems – year: 2020 ident: b37 article-title: Scholt energy control, weather data – start-page: 216 year: 2017 end-page: 220 ident: b4 article-title: Data augmentation of wearable sensor data for parkinson’s disease monitoring using convolutional neural networks publication-title: Proceedings of the 19th ACM international conference on multimodal interaction – volume: 199 year: 2021 ident: b23 article-title: An effective two-stage electricity price forecasting scheme publication-title: Electr Power Syst Res – volume: vol. 2 start-page: 2672 year: 2014 end-page: 2680 ident: b32 article-title: Generative adversarial nets publication-title: Proceedings of the 27th International conference on neural information processing systems – volume: 293 year: 2021 ident: b15 article-title: Forecasting day-ahead electricity prices: A review of state-of-the-art algorithms, best practices and an open-access benchmark publication-title: Appl Energy – volume: 211 start-page: 890 year: 2018 end-page: 903 ident: b17 article-title: Forecasting day-ahead electricity prices in europe: The importance of considering market integration publication-title: Appl Energy – year: 2017 ident: b44 article-title: Beta-VAE: Learning basic visual concepts with a constrained variational framework publication-title: ICLR – reference: Guennec AL, Malinowski S, Tavenard R. Data augmentation for time series classification using convolutional neural networks. In: ECML/PKDD workshop on advanced analytics and learning on temporal data. Riva Del Garda, Italy; 2016. – volume: vol. 27 start-page: 2672 year: 2014 end-page: 2680 ident: b12 article-title: Generative adversarial nets publication-title: Advances in neural information processing systems – volume: 24 start-page: 19 year: 2015 end-page: 36 ident: b41 article-title: Putting big data analytics to work: Feature selection for forecasting electricity prices using the LASSO and random forests publication-title: J Decis Sys – volume: 258 year: 2020 ident: b25 article-title: An adaptive hybrid model for short term electricity price forecasting publication-title: Appl Energy – year: 2016 ident: b47 article-title: Energy-based generative adversarial network – year: 2001 ident: b40 article-title: The elements of statistical learning publication-title: Springer series in statistics – start-page: 847 year: 2013 end-page: 855 ident: b43 article-title: Auto-WEKA: Combined selection and hyperparameter optimization of classification algorithms publication-title: Proceedings of the 19th ACM SIGKDD international conference on knowledge discovery and data mining – reference: Smyl S, Kuber K. Data preprocessing and augmentation for multiple short time series forecasting with recurrent neural networks. In: 36th International Symposium on Forecasting. 2016. – year: 2012 ident: b1 article-title: Learning from data – volume: 221 start-page: 386 year: 2018 end-page: 405 ident: b18 article-title: Forecasting spot electricity prices: Deep learning approaches and empirical comparison of traditional algorithms publication-title: Appl Energy – year: 2018 ident: b6 article-title: Empirical evaluation of variational autoencoders for data augmentation publication-title: VISIGRAPP – start-page: 2797 year: 2017 end-page: 2806 ident: b8 article-title: A bayesian data augmentation approach for learning deep models publication-title: Advances in neural information processing systems – start-page: 843 year: 2017 end-page: 852 ident: b2 article-title: Revisiting unreasonable effectiveness of data in deep learning era publication-title: 2017 IEEE International Conference on Computer Vision (ICCV) – volume: 11 year: 2018 ident: b21 article-title: Electricity price forecasting using recurrent neural networks publication-title: Energies – start-page: 50 year: 1947 end-page: 60 ident: b51 article-title: On a test of whether one of two random variables is stochastically larger than the other publication-title: Ann Math Stat – volume: vol. 6 start-page: 3 year: 1994 end-page: 10 ident: b11 article-title: Autoencoders, minimum description length and Helmholtz free energy publication-title: Advances in neural information processing systems – reference: Bengio Y, Mesnil G, Dauphin Y, Rifai S. Better mixing via deep representations. In: International conference on machine learning. 2013, p. 552–60. – volume: 9 start-page: 621 year: 2016 ident: b19 article-title: Automated variable selection and shrinkage for day-ahead electricity price forecasting publication-title: Energies – year: 2011 ident: b49 article-title: Principal component analysis – year: 2017 ident: b31 article-title: On unifying deep generative models – start-page: 1096 year: 2008 end-page: 1103 ident: b28 article-title: Extracting and composing robust features with denoising autoencoders publication-title: Proceedings of the 25th international conference on machine learning – year: 2017 ident: b34 article-title: Wasserstein GAN – volume: 11 start-page: 2039 year: 2018 ident: b20 article-title: Efficient forecasting of electricity spot prices with expert and LASSO models publication-title: Energies – volume: 32 start-page: 303 year: 2016 end-page: 312 ident: b9 article-title: Bagging exponential smoothing methods using STL decomposition and Box–Cox transformation publication-title: Int J Forecast – year: 2017 ident: b38 article-title: On denoising autoencoders trained to minimise binary cross-entropy – reference: Isola P, Zhu J-Y, Zhou T, Efros AA. Image-to-image translation with conditional adversarial networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2017, p. 1125–34. – volume: 1 start-page: 80 year: 1945 end-page: 83 ident: b50 article-title: Individual comparisons by ranking methods publication-title: Biom Bull – year: 2020 ident: b36 article-title: ENTSO-e – volume: 45 start-page: 5 year: 2001 end-page: 32 ident: b42 article-title: Random forests publication-title: Mach Learn – year: 2017 ident: b48 article-title: On the regularization of wasserstein GANs – year: 2020 ident: b22 article-title: Transfer learning for electricity price forecasting – volume: 9 start-page: 2044 year: 2021 end-page: 2054 ident: b26 article-title: Hybrid local general regression neural network and harmony search algorithm for electricity price forecasting publication-title: IEEE Access – year: 2013 ident: b30 article-title: Auto-encoding variational Bayes – volume: 27 start-page: 4066 year: 2018 end-page: 4079 ident: b46 article-title: Perceptual adversarial networks for image-to-image transformation publication-title: IEEE Trans Image Process – volume: 10 issue: 1 year: 2019 ident: 10.1016/j.apenergy.2021.117695_b14 article-title: Introducing technical indicators to electricity price forecasting: A feature engineering study for linear, ensemble, and deep machine learning models publication-title: Appl Sci doi: 10.3390/app10010255 – start-page: 216 year: 2017 ident: 10.1016/j.apenergy.2021.117695_b4 article-title: Data augmentation of wearable sensor data for parkinson’s disease monitoring using convolutional neural networks – volume: vol. 6 start-page: 3 year: 1994 ident: 10.1016/j.apenergy.2021.117695_b11 article-title: Autoencoders, minimum description length and Helmholtz free energy – volume: 199 year: 2021 ident: 10.1016/j.apenergy.2021.117695_b23 article-title: An effective two-stage electricity price forecasting scheme publication-title: Electr Power Syst Res doi: 10.1016/j.epsr.2021.107416 – ident: 10.1016/j.apenergy.2021.117695_b3 – start-page: 843 year: 2017 ident: 10.1016/j.apenergy.2021.117695_b2 article-title: Revisiting unreasonable effectiveness of data in deep learning era – year: 2013 ident: 10.1016/j.apenergy.2021.117695_b30 – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b44 article-title: Beta-VAE: Learning basic visual concepts with a constrained variational framework – volume: 221 start-page: 386 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b18 article-title: Forecasting spot electricity prices: Deep learning approaches and empirical comparison of traditional algorithms publication-title: Appl Energy doi: 10.1016/j.apenergy.2018.02.069 – volume: 9 start-page: 2044 year: 2021 ident: 10.1016/j.apenergy.2021.117695_b26 article-title: Hybrid local general regression neural network and harmony search algorithm for electricity price forecasting publication-title: IEEE Access doi: 10.1109/ACCESS.2020.3048519 – ident: 10.1016/j.apenergy.2021.117695_b10 – year: 2020 ident: 10.1016/j.apenergy.2021.117695_b22 – start-page: 5767 year: 2017 ident: 10.1016/j.apenergy.2021.117695_b35 article-title: Improved training of wasserstein gans – start-page: 1096 year: 2008 ident: 10.1016/j.apenergy.2021.117695_b28 article-title: Extracting and composing robust features with denoising autoencoders – volume: 1 start-page: 80 issue: 6 year: 1945 ident: 10.1016/j.apenergy.2021.117695_b50 article-title: Individual comparisons by ranking methods publication-title: Biom Bull doi: 10.2307/3001968 – year: 2016 ident: 10.1016/j.apenergy.2021.117695_b27 – year: 2011 ident: 10.1016/j.apenergy.2021.117695_b49 – volume: 11 issue: 5 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b21 article-title: Electricity price forecasting using recurrent neural networks publication-title: Energies doi: 10.3390/en11051255 – volume: 258 year: 2020 ident: 10.1016/j.apenergy.2021.117695_b25 article-title: An adaptive hybrid model for short term electricity price forecasting publication-title: Appl Energy doi: 10.1016/j.apenergy.2019.114087 – year: 2001 ident: 10.1016/j.apenergy.2021.117695_b40 article-title: The elements of statistical learning doi: 10.1007/978-0-387-21606-5 – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b34 – start-page: 2797 year: 2017 ident: 10.1016/j.apenergy.2021.117695_b8 article-title: A bayesian data augmentation approach for learning deep models – ident: 10.1016/j.apenergy.2021.117695_b29 – year: 2016 ident: 10.1016/j.apenergy.2021.117695_b47 – volume: 27 start-page: 4066 issue: 8 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b46 article-title: Perceptual adversarial networks for image-to-image transformation publication-title: IEEE Trans Image Process doi: 10.1109/TIP.2018.2836316 – year: 2018 ident: 10.1016/j.apenergy.2021.117695_b6 article-title: Empirical evaluation of variational autoencoders for data augmentation – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b5 – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b48 – start-page: 50 year: 1947 ident: 10.1016/j.apenergy.2021.117695_b51 article-title: On a test of whether one of two random variables is stochastically larger than the other publication-title: Ann Math Stat doi: 10.1214/aoms/1177730491 – volume: 45 start-page: 5 issue: 1 year: 2001 ident: 10.1016/j.apenergy.2021.117695_b42 article-title: Random forests publication-title: Mach Learn doi: 10.1023/A:1010933404324 – volume: 30 start-page: 1030 issue: 4 year: 2014 ident: 10.1016/j.apenergy.2021.117695_b16 article-title: Electricity price forecasting: A review of the state-of-the-art with a look into the future publication-title: Int J Forecast doi: 10.1016/j.ijforecast.2014.08.008 – volume: vol. 27 start-page: 2672 year: 2014 ident: 10.1016/j.apenergy.2021.117695_b12 article-title: Generative adversarial nets – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b38 – start-page: 847 year: 2013 ident: 10.1016/j.apenergy.2021.117695_b43 article-title: Auto-WEKA: Combined selection and hyperparameter optimization of classification algorithms – year: 2020 ident: 10.1016/j.apenergy.2021.117695_b36 – volume: 11 start-page: 2039 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b20 article-title: Efficient forecasting of electricity spot prices with expert and LASSO models publication-title: Energies doi: 10.3390/en11082039 – volume: 190 start-page: 291 year: 2017 ident: 10.1016/j.apenergy.2021.117695_b24 article-title: Electricity price forecasting by a hybrid model, combining wavelet transform, ARMA and kernel-based extreme learning machine methods publication-title: Appl Energy doi: 10.1016/j.apenergy.2016.12.130 – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b33 – volume: 9 start-page: 621 issue: 8 year: 2016 ident: 10.1016/j.apenergy.2021.117695_b19 article-title: Automated variable selection and shrinkage for day-ahead electricity price forecasting publication-title: Energies doi: 10.3390/en9080621 – start-page: 2535 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b7 article-title: Eeg data augmentation for emotion recognition using a conditional wasserstein gan – year: 2012 ident: 10.1016/j.apenergy.2021.117695_b1 – volume: 32 start-page: 303 year: 2016 ident: 10.1016/j.apenergy.2021.117695_b9 article-title: Bagging exponential smoothing methods using STL decomposition and Box–Cox transformation publication-title: Int J Forecast doi: 10.1016/j.ijforecast.2015.07.002 – year: 2020 ident: 10.1016/j.apenergy.2021.117695_b37 – volume: vol. 2 start-page: 2672 year: 2014 ident: 10.1016/j.apenergy.2021.117695_b32 article-title: Generative adversarial nets – year: 1998 ident: 10.1016/j.apenergy.2021.117695_b39 – volume: 211 start-page: 890 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b17 article-title: Forecasting day-ahead electricity prices in europe: The importance of considering market integration publication-title: Appl Energy doi: 10.1016/j.apenergy.2017.11.098 – volume: 293 year: 2021 ident: 10.1016/j.apenergy.2021.117695_b15 article-title: Forecasting day-ahead electricity prices: A review of state-of-the-art algorithms, best practices and an open-access benchmark publication-title: Appl Energy doi: 10.1016/j.apenergy.2021.116983 – volume: 24 start-page: 19 issue: 1 year: 2015 ident: 10.1016/j.apenergy.2021.117695_b41 article-title: Putting big data analytics to work: Feature selection for forecasting electricity prices using the LASSO and random forests publication-title: J Decis Sys doi: 10.1080/12460125.2015.994290 – volume: 321 start-page: 321 year: 2018 ident: 10.1016/j.apenergy.2021.117695_b13 article-title: Gan-based synthetic medical image augmentation for increased cnn performance in liver lesion classification publication-title: Neurocomputing doi: 10.1016/j.neucom.2018.09.013 – ident: 10.1016/j.apenergy.2021.117695_b45 doi: 10.1109/CVPR.2017.632 – year: 2017 ident: 10.1016/j.apenergy.2021.117695_b31 |
| SSID | ssj0002120 |
| Score | 2.620418 |
| Snippet | A model’s expected generalisation error is inversely proportional to its training set size. This relationship can pose a problem when modelling multivariate... |
| SourceID | proquest crossref elsevier |
| SourceType | Aggregation Database Enrichment Source Index Database Publisher |
| StartPage | 117695 |
| SubjectTerms | Adversarial network Augmentation Autoencoder case studies electricity electricity costs Electricity price forecasting energy markets Multivariate time series prediction Regression regression analysis time series analysis |
| Title | Data augmentation for time series regression: Applying transformations, autoencoders and adversarial networks to electricity price forecasting |
| URI | https://dx.doi.org/10.1016/j.apenergy.2021.117695 https://www.proquest.com/docview/2636572997 |
| Volume | 304 |
| WOSCitedRecordID | wos000701874500012&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVESC databaseName: Elsevier SD Freedom Collection Journals 2021 customDbUrl: eissn: 1872-9118 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0002120 issn: 0306-2619 databaseCode: AIEXJ dateStart: 19950101 isFulltext: true titleUrlDefault: https://www.sciencedirect.com providerName: Elsevier |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1bb9MwFLZKxwM8IBhMGzcZCfFSUtLczdsEHbdS9tBJfbMcx5myrUlo0mr9EzzyezmO7TSahjYeaKUoiuKk7ffV5zvxuSD0Ok3sKI55YkUkci1PEGFFIiJWyMM4iOHlK6Qn4XQazefkuNf7bXJh1hdhnkeXl6T8r1DDMQBbps7-A9ztReEA7APosAXYYXsr4D-ymg3Y6nShs4p0JGG2EAN5fyHXCU5V9GsT1SF16EZ3i2g1bKHkNVvVhax0KaOdVVlX2b-5Yk2rj1xFkDclIlQ3nYxLTV_KOkXypoKzqjam0VS61apXNDmHrYwWi0w1AIOpcrNsTcX3DDi5VpPRBmRq2tqHopnFvxXbTLZjJpOpz1XNhGl2Dj57UQ0-DbvPNZyRjBFRmZ0mn8sOLOnfdedq1_YGpVxoDohvXWsB1MOIsyEr1RcZymvrEVubZ9b5pz_o0clkQmfj-exN-dOS3cjkqr1uzXIH7TihT6I-2jn8Mp5_bW28owt-ms_YyT2__tZ_kz1XBECjamYP0QPtjuBDRaNHqCfyXXS_U6RyF-2Nt7mQcKo2BtVj9EsyDXeZhgF0LJmGFdPwlmnvseEZvsKzt7jLMgwswx2WYcMyXBe4wzLcsAx3WPYEnRyNZx8-W7q9h8U9n9RWKogvhHBEKFPFAjt1feGlIfjPxBlxJ_B5mnJ42zGJkljEDuEeBwEcC38kWMDdPdTPi1zsI0zsJPHcJHWIcDzuhIyDF8QiErAYPBAWHyDf_PqU69r3sgXLBTVBjmfUoEYlalShdoDeteNKVf3lxhHEgEu1hlXalAJBbxz7yrCBwiQvV-5YLopVRZ3ADXxwg0n49BbnPEP3tn-o56hfL1fiBbrL13VWLV9qKv8BwMjWFQ |
| linkProvider | Elsevier |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Data+augmentation+for+time+series+regression%3A+Applying+transformations%2C+autoencoders+and+adversarial+networks+to+electricity+price+forecasting&rft.jtitle=Applied+energy&rft.au=Demir%2C+Sumeyra&rft.au=Mincev%2C+Krystof&rft.au=Kok%2C+Koen&rft.au=Paterakis%2C+Nikolaos+G.&rft.date=2021-12-15&rft.issn=0306-2619&rft.volume=304+p.117695-&rft_id=info:doi/10.1016%2Fj.apenergy.2021.117695&rft.externalDBID=NO_FULL_TEXT |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0306-2619&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0306-2619&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0306-2619&client=summon |