The main issue concerning Business Intelligence and Analytics is how to identify critical target audience and make an effective decision to achieve the goal. This study proposes a rough set theory incorporated with logistic regression, which can evaluate the key variables more accurately and induct the important rules in particular. By using the predictive ability of logistic regression from all dataset, false positives which are the closest to the survivals of dead groups will be able to be classified. Next, the rough set theory is used to generate the reducts and induct decision rules. Through the comparison between the rules belonged to the false positives and the rule base, the key issues about the dying situation of false positives are focused and resolved. Ultimately, the cross reference decision table provides multiple policy making modes to decision makers. The household energy saving case is studied to validate the superiority of the proposed solution approach.