Optimized Random Forest Classifier Basedon Genetic Algorithm for Heart Failure Prediction

Heart failure is a serious long-term condition that usually gets worse over time. On the other hand, some people do not aware to check their heart health regularly. In this study, the Random Forest will be optimized using the Genetic Algorithm to obtain the best parameters and will be applied to the heart failure dataset from Kaggle. We experimented with two iterations for every nine combinations of the parameters. We compared the results of optimized random forest, stand-alone random forest, decisiontree, and Naïve Bayes algorithms. Our finding is that the optimized method is slightly better than the other algorithms. The best F1-score is obtained atthe second iteration which is 0.90789 compared to0.89404 obtained with the sole random forest, 0.85034 obtained with the decision tree, and 0.86195 obtained with Naive Bayes. The best recall value is 0.91925obtained in the first iteration, and in the second iteration. The best recall is also obtained with the sole random forest algorithm. The best precision valueis 0.89937 which was obtained in the first. By these results, the optimized random forest algorithm could be used to result in reliable predictions about heart failure.