Using local learning with fuzzy transform: application to short term forecasting problems

Vincenzo Loia,Stefania Tomasiello,Alfredo Vaccaro,Jinwu Gao
DOI: https://doi.org/10.1007/s10700-019-09311-x
IF: 5.274
2019-01-01
Fuzzy Optimization and Decision Making
Abstract:In this paper, we formally discuss a computational scheme, which combines a local weighted regression model with fuzzy transform (or F-transform for short). The latter acts as a reduction technique on the cardinality of the learning problem, resulting in a more efficient algorithm. We tested the proposed approach first through two typical benchmark problems, that is the Hénon and the Mackey–Glass chaotic time series, then we applied it to short-term forecasting problems. Short-term forecasting is important in the energy field for the management of power systems and for energy trading. Hence, we considered two typical application examples in this field, that is wind power forecasting and load forecasting. Numerical results show the effectiveness of the proposed approach through a comparison against alternative techniques.
What problem does this paper attempt to address?