Black hole algorithm as a heuristic approach for rare event classification problem

The logistic regression is generally preferred when there is no big difference in the occurrence frequencies of two possible results for the considered event. However, for the events occurring rarely such as wars, economic crisis and natural disasters, namely having relatively small occurrence frequency when compared to the general events, the logistic regression gives biased parameter estimations. Therefore, the logistic regression underestimates the occurrence probability of the rare events. In this study, black hole algorithm is proposed and used to obtain unbiased estimation parameters for rare events, instead of using the classical logistic regression approach. In order to estimate the logistic regression parameter for the cases dichotomous event groups are rare, we propose a black hole algorithm (BHA) approach. For the samples with different rareness degrees, we obtain the parameter values and their bias and root mean square errors for BHA and logistic regression, and then compare them. Moreover, we also investigate the classification performance of two methods on a real life data. As a result, we obtained that BHA gives less biased estimates in simulation and real-life data compared to logistic regression.