CREATION OF SOFTWARE FOR SIMULATION OF SEASONALITY IN TIME SERIES
DOI:
https://doi.org/10.31891/2307-5732-2024-343-6-14Keywords:
time series, forecasting, seasonality, additive model, multiplicative modelAbstract
The main objective of this work is to create a software tool for generating forecasts for processes that can be described by time series and that have a seasonal effect based on additive and multiplicative models We analyzed processes predictable by these models. At the outset, modeling was undertaken in Microsoft Excel. The built model is used as a control example to check the correctness of the developed software. To begin with, this time series was analyzed for variability using the coefficient of variation. Since its value exceeds 33%, namely 36%, a multiplicative time series model was chosen for forecasting.
Based on the analysis of the visual presentation of changes in the measured values of the farm's profit, it can be concluded that the data of the series indicate the presence of a pronounced trend and seasonal effect, which indicates an increase in income in the 4th quarter. The work describes step-by-step instructions for building a model and building a forecast. The trend parameters were found using the least squares method. To build a forecast, we multiply the trend by the seasonal component. Analysis of the generated forecast graph demonstrates the model's ability to replicate the dynamics of the input time series.. The adequacy of the built model is confirmed by the coefficient of determination, the value of which is 96,32%.
To automate forecast calculations in the Visual Studio Community 2022 environment, using the additional .NET Framework platform, a forecast calculation window program based on the additive and multiplicative time series model for Windows has been developed. C# was utilized to develop. An initial view of the functionality of the program and a form illustrating the final demonstration of the capabilities of the developed software are provided. A forecast and a confidence interval for the forecast for the next quarter have been constructed.
The capabilities of the ready-to-use developed software product are demonstrated.
