Estimated gross domestic product Turkey's with dynamic linear models

The economic forecast always plays an important role as a decision-making tool when formulating financial policy and strategy for businesses and governments. However, when dealing with real data (not just about economics, but also sociology, biology, engineering, etc.), we rarely know enough about the topic of interest. Even if there is a precise and deterministic model that defines the system we want to analyze, it is always possible to skip some explanatory variables or make measurement errors. Thus, an amount of uncertainty will always be available when analyzing the subject being investigated. Therefore, we cannot see the estimates made as a definitive science. In recent years, there has been an increasing interest in the application of state space models (DUM) for time series analysis purposes. The models of this category describe a specific time series as a dynamic system and a random component of disorder. Its powerful probability structures offer a flexible framework that can be used for a wide range of applications. This flexibility causes us to choose a subclass of DUMs, ie, dynamic linear models (DLM). One of the important variables that show the economics strength of the countries in the globalized world is the gross domestic product (GDP). GDP values of countries with a strong economy are high. In underdeveloped countries, GDP figures are generally low. In this thesis, a subclass of DUMs, DLMs are explained and their use in time series is introduced. With dynamic linear models using GDP data of Turkey are making predictions for the coming years. For this purpose, the R program, which is very useful and useful in recent years, is used for applications and predictions. Key Words: Dynamic Linear Models (DLM), State Space Models (DUM), Time Series, Stationarity, Gross Domestic Product