TR2009-050

Memory-Based Modeling of Seasonality for Prediction of Climatic Time Series


    •  Nikovski, D.N.; Ramachandran, G., "Memory-Based Modeling of Seasonality for Prediction of Climatic Time Series", International Conference on Machine Learning and Data Mining in Pattern Recognion (MLDM), ISSN: 0302-9743, July 2009, vol. 5632, pp. 734-748.
      BibTeX Download PDF
      • @inproceedings{Nikovski2009jul1,
      • author = {Nikovski, D.N. and Ramachandran, G.},
      • title = {Memory-Based Modeling of Seasonality for Prediction of Climatic Time Series},
      • booktitle = {International Conference on Machine Learning and Data Mining in Pattern Recognion (MLDM)},
      • year = 2009,
      • volume = 5632,
      • pages = {734--748},
      • month = jul,
      • issn = {0302-9743},
      • url = {http://www.merl.com/publications/TR2009-050}
      • }
  • MERL Contact:
  • Research Area:

    Data Analytics


TR Image

The paper describes a method for predicting climate time series that consist of significant annual and diurnal seasonal components and a short-term stockastic component. A memory-based method for modeling of the non-linear seasonal components is proposed that allows the application of simpler linear models for predicting short-term deviations from seasonal averages. The proposed method results in significant reduction of prediction error when predicting error time series of ambient air temperature from multiple locations. Moreover, combining the statistical predictor with meteorological forecasts using linear regression or Kalman filtering further reduces error to typically between 1°C over a prediction horizon of one hour and 2.5°C over 24 hours.