APLIKASI SISTEM PAKAR UNTUK MEMPREDIKSI HASIL PERTANDINGAN SEPAK BOLA MENGGUNAKAN DEMPSTER-SHAFER THEORY (DST)

Abstract

An expert system application for football match results prediction with Dempster-Shafer Theory (DST) which can predict victory, draw, or lose chances from a competing team in a football match. Predictions are performed based on factors which are home-away factor, match history, last match intervals of each team, numbers of absent key players, match motivation, players' abilities, coach presence, and betting odds for each team. Collected data will be processed to data that will be used for basic probability assignment (bpa) value calculation. The processed bpa value will be used for inference process based on Dempster-Shafer Theory (DST) calculation. The result is the highest belief value which will be used as a value for evidence probability calculation that resulted in belief value for lower limit and plausibility value for upper limit. The output which will be displayed are victory chance interval, draw chance, or lose chance interval for the competing team. From the test case which consists of 100 football match data, it can be inferred that the prediction accuracy is 62% based on factors described earlier