Türkiye'de Gün Öncesi Piyasası İçin Elektrik Fiyatlarının Tahmini
![](/xmlui/themes/Mirage2//images/mime2.png)
Göster/ Aç
Tarih
2020Yazar
Demirezen, Sinan
Ambargo Süresi
Acik erisimÜst veri
Tüm öğe kaydını gösterÖzet
Electricity, known as static electricity from ancient times to 1880’s, take a indispensable place in human life especially down from it had been used as a lightening purpose. At the present time, there isn’ t any substitute goods with it. So, electricity is one of the most important ones daily life. İn the first periods of active using electricity, states undertook its production, transmision and distribution mechanism via establishing their institution. Particularly, while it was preferred electricity industry under the state monopoly in 1940’s, it was began to refuse this thought due to the inefficiency and deactivation in the industry in 1980’s. In this period, so that states had regulatory and supervisory position, in the electricity industry was began to regulate by reducing its authority. Primarily Chile, afterwards England, Nordic countries and many other countries regulated their industries in scope of liberalization. Therefore, electricity markets which is also involve private sector’s agent was established. In Turkey, liberalization of electricity market was began at the beginning of 2000’s. İn the aim of this master’ s thesis was designed as prediction and forecasting of electricity prices. In the scope of this thesis consist of Market Clearance Prices (MCP) which is defined as a result of making bid by market participant in Day Ahead Market. Artificial Neural Network, Support Vector Machine, Random Forest including among Machine Learning methods used for prediction and forecasting of the prices. On the other hands, MCP was tackled with six different models consist of named as Model 1, Model 2, Model 3, Model 4, Model 5 and Model 6. The method and model which is the best of prediction performance among aforementioned methods and models was used for forecasting electricity prices. As a result of the analysis, the best method and the best model to predict MCP is random forest and Model 6, respectively. From this point of wiew, twenty four hourly MCP was forecasted with Model 6. Therefore, forecasting results will give an idea (to market operator, market participant and individuals to study in this area) about prices and the methods using in this paper.