t-SNE를 이용한 전력수요 데이터의 군집분석
Cluster Analysis of Daily Electricity Demand with t-SNE
0
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
For an efficient management of electricity market and power systems, accurate forecasts for electricity demand are essential. Since there are many factors, either known or unknown, determining the realized loads, it is difficult to forecast the demands with the past time series only. In this paper we perform a cluster analysis on electricity demand data collected from Jan. 2000 to Dec. 2017. Our purpose of clustering on electricity demand data is that each cluster is expected to consist of data whose latent variables are same or similar values. Then, if properly clustered, it is possible to develop an accurate forecasting model for each cluster separately. To validate the feasibility of this approach for building better forecasting models, we clustered data with t-SNE. To apply t-SNE to time series data effectively, we adopt the dynamic time warping as a similarity measure. From the result of experiments, we found that several clusters are well observed and each cluster can be interpreted as a mix of well-known factors such as trends, seasonality and holiday effects and other unknown factors. These findings can motivate the approaches which build forecasting models with respect to each cluster independently.
Abstract
For an efficient management of electricity market and power systems, accurate forecasts for electricity demand are essential. Since there are many factors, either known or unknown, determining the realized loads, it is difficult to forecast the demands with the past time series only. In this paper we perform a cluster analysis on electricity demand data collected from Jan. 2000 to Dec. 2017. Our purpose of clustering on electricity demand data is that each cluster is expected to consist of data whose latent variables are same or similar values. Then, if properly clustered, it is possible to develop an accurate forecasting model for each cluster separately. To validate the feasibility of this approach for building better forecasting models, we clustered data with t-SNE. To apply t-SNE to time series data effectively, we adopt the dynamic time warping as a similarity measure. From the result of experiments, we found that several clusters are well observed and each cluster can be interpreted as a mix of well-known factors such as trends, seasonality and holiday effects and other unknown factors. These findings can motivate the approaches which build forecasting models with respect to each cluster independently.
참고문헌 (10)
- "Korea Power Exchange"[URL]
- P.J. Brockwell., "Introduction to Time Series and Forecasting" (2016), Springer
- 박다인, "시간단위 전력사용량 시계열 패턴의 군집 및 분류분석", 한국데이터정보과학회지, 28(2), pp.395~406 (2017), 한국데이터정보과학회[DOI]
- 임종훈, "단기 전력수요예측 정확도 개선을 위한 대표기온 산정방안", 조명.전기설비학회논문지, 27(6), pp.39~43 (2013), 한국조명.전기설비학회
- 윤상후, "시간단위 전력수요자료의 함수적 군집분석: 사례연구", 한국데이터정보과학회지, 26(4), pp.885~894 (2015), 한국데이터정보과학회[DOI]
- L.J.P. van der Maaten, "Visualizing high-dimensional data using t-SNE", Journal of Machine Learning Research, 9, pp.2579~2695 (2008)
- G.E. Hinton., "Stochastic neighbor embedding", Proceedings of Advances in Neural Information Processing Systems (NIPS), pp.833~840 (2002)
- J.Kruskal., "The symmetric time warping problem: From continuous to discrete", Proceedings of Time Waprs, String Edits and Macromolecules: The Theory and Practice of Sequence Comparison, pp.125~161 (1983)
- S. Salvador., "Toward accurate dynamic time warping in linear time and space", Intelligent Data Analysis, 11(5), pp.561~580 (2007)
- N.V. Prasad., "Improved cepstral mean and variance normalization using Bayesian framework", Proceedings of IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU), pp.156~161 (2013)