The prediction Brahman X Charolais and Brahman X Thai native with Evolutionary-Extreme Learning Machine in Phetchabun Province Thailand.

บทคัดย่อ

The purpose of this study was to develop a growth forecasting model for Stocker Cattle Producers. In the long yearling (12-24 months) breed, only Brahman X Charolais and Brahman X Thai native were selected, all of which are 50 % breed (F1) from 17 Sub-Districts in Muang District, Phetchabun, Thailand. There were three different forecasting model methods: 1) Extreme Learning Machine (ELM); 2) Weighted- Extreme Learning Machine (Weighted ELM) and 3) Evolutionary- Extreme Learning Machine (ELM). The surveyed data were taken through the Feature Selection process to select factors affecting the relationship between feed intake, along with the growth rate according to body weight during the month. The recording data was then used to create a forecasting model to create an estimate of the weight of beef cattle that is close to the best growth. The results indicated that Evolutionary-ELM was able to gain weight during the 19th months on an average of up to 500 kilograms after the cattle were fed continuously until the 24th months. The prognostic results of Brahman X Charolais beef cattle could be predicted as an average weight of 604.088 kilograms an error at R2=0.9327, with growth rate Y=30.562x-129.40, RMSE=0.130, MSE=0.107. Brahman X Thai native beef cattle could be predicted for 509.982 kilograms an error at R2=0.9706, with growth rate Y=28.098x-164.37, RMSE=0.117, MSE=0.083. The Evolutionary-ELM algorithm learns and adjusts the weights to optimize the best results for each month's growth period, which reduces the problem of margin, keeping the distance not too high or too low. It can be used as a model for raising large beef cattle weighing 550-650 kilograms, allowing farmers who are interested in beef cattle farming to compare the model with their decision-making.