6 results
Search Results
Now showing 1 - 6 of 6
Article Citation - WoS: 16Citation - Scopus: 20Forecasting Air Quality in Tripoli: an Evaluation of Deep Learning Models for Hourly Pm2.5 Surface Mass Concentrations(Mdpi, 2023) Esager, Marwa Winis Misbah; Unlu, Kamil DemirberkIn this article, we aimed to study the forecasting of hourly PM2.5 surface mass concentrations in the city of Tripoli, Libya. We employed three state-of-the-art deep learning models, namely long short-term memory, gated recurrent unit, and convolutional neural networks, to forecast PM2.5 levels using univariate time series methodology. Our results revealed that the convolutional neural networks model performed the best, with a coefficient of variation of 99% and a mean absolute percentage error of 0.04. These findings provide valuable insights into the use of deep learning models for forecasting PM2.5 and can inform decision-making regarding air quality management in the city of Tripoli.Article Citation - WoS: 16Citation - Scopus: 18Modeling and Forecasting of Monthly Pm2.5 Emission of Paris by Periodogram-Based Time Series Methodology(Springer, 2021) Akdi, Yilmaz; Golveren, Elif; Unlu, Kamil Demirberk; Yucel, Mustafa ErayIn this study, monthly particulate matter (PM2.5) of Paris for the period between January 2000 and December 2019 is investigated by utilizing a periodogram-based time series methodology. The main contribution of the study is modeling the PM2.5 of Paris by extracting the information purely from the examined time series data, where proposed model implicitly captures the effects of other factors, as all their periodic and seasonal effects reside in the air pollution data. Periodicity can be defined as the patterns embedded in the data other than seasonality, and it is crucial to understand the underlying periodic dynamics of air pollutants to better fight pollution. The method we use successfully captures and accounts for the periodicities, which could otherwise be mixed with seasonality under an alternative methodology. Upon the unit root test based on periodograms, it is revealed that the investigated data has periodicities of 1 year and 20 years, so harmonic regression is utilized as an alternative to Box-Jenkins methodology. As the harmonic regression displayed a better performance both in and out-of-sample forecasts, it can be considered as a powerful alternative to model and forecast time series with a periodic structure.Article Citation - WoS: 1Citation - Scopus: 1Wavelet-Enhanced Sequence-To Modeling With Attention Mechanism for Short-Term Wind Power Forecasting(Taylor & Francis inc, 2025) Karaca, Burak; Unlu, Kamil Demirberk; Turkan, SemraElectricity load forecasting is crucial to managing electric systems, especially loads produced from renewable energy sources since the load from renewable energy sources varies when compared with nonrenewable sources. Turkey is producing an increasing amount of electricity from wind energy every day. The aim of this study is to introduce a hybrid deep learning model based on sequence-to-sequence learning (seq-2-seq), attention mechanisms, and wavelet transformation. Long Short-Term Memory (LSTM), Gated Recurrent Unit, and Bidirectional Long Short-Term Memory (BiLSTM) are used as decoders and encoders in the seq-2-seq model. We proposed six different models. All models are univariate type, requiring only the data itself. The model can be used on any wind farms without requiring the meteorological data. We test the proposed model on four different wind farms in Turkey: Soma, Biga, Balikesir, and Mersin. We utilize four different performance metrics to test the model's performance: mean squared error (MSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determinations (R2). The best model is seen as Wavelet-Seq2Seq-BiLSTM-LSTM at Biga Wind Farm, which achieved the best performance with a MAE of 0.127, an MSE of 0.001, a MAPE of 0.28, and an R2 of 0.997.Article Citation - WoS: 12Citation - Scopus: 13A two-step machine learning approach to predict S&P 500 bubbles(Taylor & Francis Ltd, 2021) Kabran, Fatma Basoglu; Unlu, Kamil DemirberkIn this paper, we are interested in predicting the bubbles in the S&P 500 stock market with a two-step machine learning approach that employs a real-time bubble detection test and support vector machine (SVM). SVM as a nonparametric binary classification technique is already a widely used method in financial time series forecasting. In the literature, a bubble is often defined as a situation where the asset price exceeds its fundamental value. As one of the early warning signals, prediction of bubbles is vital for policymakers and regulators who are responsible to take preemptive measures against the future crises. Therefore, many attempts have been made to understand the main factors in bubble formation and to predict them in their earlier phases. Our analysis consists of two steps. The first step is to identify the bubbles in the S&P 500 index using a widely recognized right-tailed unit root test. Then, SVM is employed to predict the bubbles by macroeconomic indicators. Also, we compare SVM with different supervised learning algorithms by usingk-fold cross-validation. The experimental results show that the proposed approach with high predictive power could be a favourable alternative in bubble prediction.Article Citation - WoS: 18Citation - Scopus: 20Periodicity in Precipitation and Temperature for Monthly Data of Turkey(Springer Wien, 2021) Akdi, Yilmaz; Unlu, Kamil DemirberkIn this study, we model and forecast monthly average temperature and monthly average precipitation of Turkey by employing periodogram-based time series methodology. We compare autoregressive integrated moving average methodology and harmonic regression. We show that harmonic regression performs better than the classical methodology in both time series. Also, we find that the monthly average temperature and monthly average precipitation have two different periodic structures of 6 months and 12 months which coincide with the seasonal pattern of the time series.Article Citation - WoS: 9Citation - Scopus: 7Identifying the Cycles in Covid-19 Infection: the Case of Turkey(Taylor & Francis Ltd, 2023) Akdi, Yilmaz; Karamanoglu, Yunus Emre; Unlu, Kamil Demirberk; Bas, CemThe new coronavirus disease, called COVID-19, has spread extremely quickly to more than 200 countries since its detection in December 2019 in China. COVID-19 marks the return of a very old and familiar enemy. Throughout human history, disasters such as earthquakes, volcanic eruptions and even wars have not caused more human losses than lethal diseases, which are caused by viruses, bacteria and parasites. The first COVID-19 case was detected in Turkey on 12 March 2020 and researchers have since then attempted to examine periodicity in the number of daily new cases. One of the most curious questions in the pandemic process that affects the whole world is whether there will be a second wave. Such questions can be answered by examining any periodicities in the series of daily cases. Periodic series are frequently seen in many disciplines. An important method based on harmonic regression is the focus of the study. The main aim of this study is to identify the hidden periodic structure of the daily infected cases. Infected case of Turkey is analyzed by using periodogram-based methodology. Our results revealed that there are 4, 5 and 62 days cycles in the daily new cases of Turkey.
