A Tensor-Based Method for Geosensor Data Forecasting
In recent years, geosensor data forecasting has received considerable attention. However, the presence of correlation (i.e. spatial correlation across several sites and time correlation within each site) poses difficulties to accurate forecasting. In this paper, a tensor-based method for geosensor data forecasting is proposed. Specifically, a tensor pattern is first introduced into modelling the geosensor data, which can take advantage of geosensor spatial-temporal information and preserve the multi-way nature of geosensor data, and then a tensor decomposition based algorithm is developed to forecast future values of time series. The proposed approach not only combines and utilizes the multi-mode correlations, but also well extracts the underlying factors in each mode of tensor and mines the multi-dimensional structures of geosensor data. Experimental evaluations on real world geosensor data validate the effectiveness of the proposed methods.
KeywordsGeosensor data forecasting Tensor decomposition CP-WOPT model
This research was supported by the National Natural Science Foundation of China (61762090, 61262069, 61472346, and 61662086), The Natural Science Foundation of Yunnan Province (2016FA026, 2015FB114), the Project of Innovative Research Team of Yunnan Province, and Program for Innovation Research Team (in Science and Technology) in University of Yunnan Province (IRTSTYN).
- 1.Yang, B., Guo, C., Jensen, C.S.: Travel cost inference from sparse, spatio temporally correlated time series using Markov models. PVLDB 6(9), 769–780 (2013)Google Scholar
- 2.Yu, R., Cheng, D., Liu, Y.: Accelerated online low rank tensor learning for multivariate spatiotemporal streams. ICML 2015, 238–247 (2015)Google Scholar
- 4.Pravilovic, S., Appice, A., Malerba, D.: An intelligent technique for forecasting spatially correlated time series. In: Baldoni, M., Baroglio, C., Boella, G., Micalizio, R. (eds.) AI*IA 2013. LNCS (LNAI), vol. 8249, pp. 457–468. Springer, Cham (2013). https://doi.org/10.1007/978-3-319-03524-6_39CrossRefGoogle Scholar
- 5.Pravilovic, S., Appice, A., Malerba, D.: Integrating cluster analysis to the ARIMA model for forecasting geosensor data. In: Andreasen, T., Christiansen, H., Cubero, J.-C., Raś, Zbigniew W. (eds.) ISMIS 2014. LNCS (LNAI), vol. 8502, pp. 234–243. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-08326-1_24CrossRefGoogle Scholar
- 11.Pokrajac, D., Obradovic, Z.: Improved spatial-temporal forecasting through modelling of spatial residuals in recent history. In: SDM, Chicago, IL, USA, 5–7 April 2001, pp. 1–17 (2001)Google Scholar
- 13.Ohashi, O., Torgo, L.: Wind speed forecasting using spatio-temporal indicators. In: ECAI, France, 27–31 August 2012, pp. 975–980 (2012)Google Scholar
- 14.Asteriou D., Hall S.: ARIMA models and the box-jenkins methodology. In: Applied Econometrics, 2nd edn., pp. 265–286. Palgrave MacMillan (2011)Google Scholar
- 16.Tucker, L.R.: Implications of factor analysis of three-way matrices for measurement of change. In: Harris, C.W. (ed.) Problems in Measuring Change, pp. 122–137. University of Wisconsin Press (1963)Google Scholar