Oeconomica Jadertina, Vol. 10 No. 2, 2020.
Izvorni znanstveni članak
Modelling and forecasting the number of employees in tourism and hotel industry in the Republic of Croatia using the model of artificial neural networks
Tea Baldigara
orcid.org/0000-0003-2563-2533
; Fakultet za menadžment u turizmu i ugostiteljstvuSveučilište u RijeciNaselje Ika, Primorska 42, OpatijaHrvatska
Sažetak
The paper investigates the performance and prognostic power of artificial neural network models in modelling and forecasting of time series of seasonal character. Models of artificial neural networks have been applied in modelling and forecasting the monthly total number of employees, the number of employed men and the number of employed women in the activity of providing accommodation services and preparing and serving food and beverages in the Republic of Croatia. The obtained modelling results have been compared with the results obtained by applying some of the traditionally used quantitative models in the analysis of seasonal time series, such as the Holt-Winters model of triple exponential smoothing and the seasonal multiplicative model of exponential trend. The evaluation of the performance and prognostic power of individual models was performed by comparing the average absolute and average absolute percentage error and the correlation coefficient between the actual and estimated values, and the predicted values were compared with the actual values. The evaluation of the obtained results showed that the selected model of acyclic multilayer perceptron is suitable for modelling and forecasting time series of seasonal character. The comparison of prognostic powers and actual and projected values of the number of employees suggests that the designed model of the artificial neural network is very reliable. This indicates that the models of artificial neural networks have great application potentials in the domain of modelling and forecasting of time series of a seasonal character.
Ključne riječi
number of employees; tourism and hotel industry; seasonal time series; artificial neural network models; multilayer perceptron model
Hrčak ID:
249817
URI
Datum izdavanja:
17.12.2020.
Posjeta: 906 *