TIME SERIES FORECASTING MODELS IN INVESTMENT PROJECTS

The article examines modern approaches to time series forecasting in investment projects, which serves as a key tool for making informed decisions under economic uncertainty. The author explores various forecasting models, including statistical (regression, autoregressive, moving average models, exponential smoothing models), structural (neural networks, Markov chains, classification and regression trees, fuzzy logic), and combined models that integrate the advantages of multiple approaches. Each of these models has its own features, advantages, and limitations that must be considered when evaluating investment projects. The paper emphasizes that the correct choice of forecasting methodology depends on the specifics of the investment project, its industry, duration, and level of uncertainty. Statistical models are effective for simpler forecasts with linear relationships, while structural and combined models are better suited for more complex tasks that account for nonlinearity and the influence of numerous factors. For example, neural network models have the ability to model complex interrelationships between different project parameters, making them particularly useful for long-term forecasts and complex investment decisions. One of the key advantages of combined models is the ability to merge the strengths of different approaches to achieve more accurate results. This is especially important in rapidly changing markets, where flexibility and adaptability to new conditions are critically important for successful investment management. For instance, combined models based on ARIMA and neural networks allow for more effective forecasting of economic indicators under high instability. The author also highlights the importance of properly configuring model parameters, as forecast accuracy depends on this. Key factors include not only the choice of model but also the volume and quality of available data, as well as the level of market turbulence. Time series forecasting is highlighted as a crucial tool for managing investment projects, especially under conditions of high risk and uncertainty. Selecting the right model, adapted to the specific conditions of a project, helps to reduce risks and improve the efficiency of managerial decisions.