Forecasting prices of broiler chickens using the creeping trend
2018
Zielinska-Sitkiewicz, M., Warsaw Univ. of Life Sciences (Poland) | Chrzanowska, M., Warsaw Univ. of Life Sciences (Poland)
The article raises the issue of forecasting prices of broiler chickens. The forecasts were generated on a set of the weekly time series of mentioned prices in the 2011−2017 period. The forecasting methods which were used in this research are adaptive methods: simple random walk model and creeping trend with fixed segments of linear trends equal 5, 7, 9 and 11 periods. The accuracy of forecasts was verified in retrospect by preparing forecast in the past, forecasting errors and graphical analysis. Both the crawling trend model and the random walk model with greater weight take into account observations closer to the forecasted values, which worked well in the case of fairly large distortions of random variations in a series of purchase prices for broiler chickens. Reducing the length of the segment in case of large random fluctuations and breakdowns of the trend allows to obtain smaller forecast errors.
显示更多 [+] 显示较少 [-]