Bounded affine barycentric-coordinate prediction for improving local extreme value forecasting

Time series data from complex systems, such as floods and economic systems, are nonlinear. Predicting local maxima and minima is important in real-world applications. However, many local state-space prediction methods represent future states as convex combinations of neighboring observations. As a consequence, predictions are effectively restricted to interpolation within the local neighborhood, which can lead to the underestimation of local extreme values. We propose an extrapolative method of time-series prediction based on barycentric coordinates. We apply the proposed method to two chaotic toy models and to an exploratory benchmark using real-world flood data. We evaluate its performance in both recursive and direct prediction. For the clean R"{o}ssler and Lorenz'63 data, the affine method yielded positive median improvements in recursive local-extreme prediction, although the mean improvement was often negative. Under observation noise, most mean and median improvements were negative, and direct prediction was generally unstable. These results indicate that the benefit of bounded affine extrapolation is restricted to selected clean-data and short- to moderate-horizon conditions.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-19
DOI
https://doi.org/10.21203/rs.3.rs-11587794/v1
Primary Topic
Time Series Analysis and Forecasting
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Bounded affine barycentric-coordinate prediction for improving local extreme value forecasting

Tohru Kawabe, Naoki Fujiwara, Yoshito Hirata
Research Square
Time Series Analysis and Forecasting
preprint

Bounded affine barycentric-coordinate prediction for improving local extreme value forecasting

Tohru Kawabe, Naoki Fujiwara, Yoshito Hirata
preprint en

Abstract

Time series data from complex systems, such as floods and economic systems, are nonlinear. Predicting local maxima and minima is important in real-world applications. However, many local state-space prediction methods represent future states as convex combinations of neighboring observations. As a consequence, predictions are effectively restricted to interpolation within the local neighborhood, which can lead to the underestimation of local extreme values. We propose an extrapolative method of time-series prediction based on barycentric coordinates. We apply the proposed method to two chaotic toy models and to an exploratory benchmark using real-world flood data. We evaluate its performance in both recursive and direct prediction. For the clean R\"{o}ssler and Lorenz'63 data, the affine method yielded positive median improvements in recursive local-extreme prediction, although the mean improvement was often negative. Under observation noise, most mean and median improvements were negative, and direct prediction was generally unstable. These results indicate that the benefit of bounded affine extrapolation is restricted to selected clean-data and short- to moderate-horizon conditions.

Research Square
University of Tsukuba (JP)
Time Series Analysis and Forecasting
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon