Forecasting models of agricultural process based on fuzzy time series

Abstract

Today, the problem of increasing the validity and accuracy of forecasts based on the analysis of time series under conditions of uncertainty is very important. The models and methods used to predict the dynamics of agricultural processes are built on quantitative information and are implemented as part of a statistical approach. In this approach, time-based forecasting models are constructed on several requirements for the initial data, the main of which are the requirements of comparability, sufficient representativeness to reveal regularity, uniformity, and stability. Only keeping these requirements, uncertainty can be interpreted in terms of randomness and appropriate statistical forecasting methods can be applied. However, the real dynamic processes taking place in agriculture are represented by time series, for which these requirements are rarely feasible, due to the great uncertainty of the factors determining their dynamics. The problem of forecasting such series is particularly relevant for agricultural science and practice. The article touches upon the possibilities of using fuzzy modeling tools to predict the dynamics of processes in the agricultural sector.