One of the weaknesses of the Bayesian Network is that it is difficult to get agreement from some experts, for an expert it will be difficult to determine the probability value, and an expert will take a long time just to build the Bayesian Network structure. To overcome the weakness of the Bayesian Network then required another artificial intelligence science that is data mining with Association Rule technique using FP-Growth algorithm. This research takes the case on eye disease with the aim of building a Bayesian Network structure and generating probability values to get where the most influential symptoms in eye disease. The method test is carried out by using data mining tools WEKA 3.7.10, with the results obtained by 24 rules that meet the provisions and qualitative test results of 99% correct and get the probability value for presbyopia disease, with the greatest influence on women, evidenced by probability value of 60 %. For the most influential age was the mature middle of 31-59 years at 65%, and the most influential symptom was a near blur of 98%. As for conjunctivitis disease with the largest influence on men by 53%.For the most influential age is the middle adult from 31-59 years by 43%, and the most influential symptoms are the sticky eye of 100%. Based on the results of these tests can be concluded that the Association Rule technique succeeded in overcoming Bayesian Network weaknesses based on facts and data.