Browsing by Author "Ozdes, Celik"

Now showing 1 - 2 of 2

Forecasting Critical Economic & Political Events Via Electricity Consumption Patterns in the United States of America and Turkey
(Springernature, 2025) Ozdes, Celik; Ediger, Volkan S.; Eroglu, Deniz
Impacts from natural disasters, government decisions and public's reactions can significantly alter societal daily routines. These effects resonate in systems where individual contributions, such as energy consumption, serve as indirect indicators of societal welfare and living standards. Preparedness for unforeseen events is crucial to enhancing societal well-being. Thus, analysing historical data for unexpected critical transitions and forecasting future occurrences is paramount. Recurrence properties of gross monthly electricity consumption in the United States of America and Turkey are examined, revealing coinciding critical periods with extreme regimes identified by a determinism time series. An ensemble of neural network proxies is then employed to forecast critical periods within a limited time frame, enabling the anticipation of similar occurrences. Validation of this approach demonstrates high predictive performance when measured quantities adequately reflect underlying system dynamics. Predictions based on electricity consumption data suggest potential systemic and socioeconomic crises for both nations within one year, with probabilities, 85% for the US and 32% for Turkey.
Citation - WoS: 6
Citation - Scopus: 6
Transformation Cost Spectrum for Irregularly Sampled Time Series
(Springer Heidelberg, 2023) Ozdes, Celik; Eroglu, Deniz; Molecular Biology and Genetics; 05. Faculty of Engineering and Natural Sciences; 01. Kadir Has University
Irregularly sampled time series analysis is a common problem in various disciplines. Since conventional methods are not directly applicable to irregularly sampled time series, a common interpolation approach is used; however, this causes data distortion and consequently biases further analyses. We propose a method that yields a regularly sampled time series spectrum of costs with minimum information loss. Each time series in this spectrum is a stationary series and acts as a difference filter. The transformation costs approach derives the differences between consecutive and arbitrarily sized segments. After obtaining regular sampling, recurrence plot analysis is performed to distinguish regime transitions. The approach is applied to a prototypical model to validate its performance and to different palaeoclimate proxy data sets located around Africa to identify critical climate transition periods during the last 5 million years and their characteristic properties.