A Time Series Forecast Model Based on the Most Similar Pattern
Year: 2026
Type: article
Abstract: The paper presents a time series forecasting model based on the most similar pattern (FMMSP), which identifies historically correlated segments to forecast future values without requiring time series stationarity. The model is validated on electricity consumption and day-ahead price data from Russian wholesale market, achieving mean absolute percentage errors as low as 1.04 % for consumption and 7.0 % for prices.
Source: INFORMACIONNYE TEHNOLOGII
Author: Chuchueva, Irina
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
SDG: Climate action
Open Access status: hybrid
OpenAlex ID: W7156349283
FWCI:
Cites:
Cited by:
Related to: