Bootstrap Causal Feature Selection for irrelevant feature elimination

Irrelevant features may lead to degradation in accuracy and efficiency of classifier performance. In this paper, Bootstrap Causal Feature Selection (BCFS) algorithm is proposed. BCFS uses bootstrapping with a causal discovery algorithm to remove irrelevant features. The results are evaluated by the number of selected features and classification accuracy. According to the experimental results, BCFS is able to remove irrelevant features and provides slightly higher average accuracy than usingIrrelevant features may lead to degradation in accuracy and efficiency of classifier performance. In this paper, Bootstrap Causal Feature Selection (BCFS) algorithm is proposed. BCFS uses bootstrapping with a causal discovery algorithm to remove irrelevant features. The results are evaluated by the number of selected features and classification accuracy. According to the experimental results, BCFS is able to remove irrelevant features and provides slightly higher average accuracy than using the original features and causal feature selection. Moreover, BCFS also reduces complexity in causal graphs which provides more comprehensibility for the casual discovery system. the original features and causal feature selection. Moreover, BCFS also reduces complexity in causal graphs which provides more comprehensibility for the casual discovery system.