Skip to main content

InfoTech conference

2020 IEEE International Conference on Information Technologies

Multiobjective Forecasting Model Based on the Interval Discrete Type-2 Fuzzy Sets and Genetic Algorithm

Liliya Anatolievna Demidova
Maksim Anatolievich Stepanov
Russian Technological University ‚Äď MIREA, Moscow
Ryazan State Radio Engineering University
Russian Federation

The forecasting task for the short time series has been investigated. The main problem of this task is the availability of a small amount of relevant information. Such tasks can be considered as the tasks in the conditions of uncertainty, fuzziness of the initial data. We suggest to develop the multiobjective forecasting model based on the interval discrete type-2 fuzzy sets. As the criteria which should be minimized, it is proposed to use the average relative forecasting error and the tendency mismatch indicator. The search for the optimal parameters values of the forecasting model is implemented using the genetic algorithm. The examples showing the possibilities of the proposed forecasting model have been discussed.

Key words:
short time series
type-2 fuzzy set
average relative forecasting error