Specific boosting method for scientific discovery

In this paper we present a specific boosting method for decision trees (SBMDT) in the field of scientific discovery. Ensemble classification methods have been shown to lead to reduce classification error on unseen cases. However, a high accuracy ratio of prediction sometimes requires complicated predictors, and makes it hard to understand the simple laws affecting the values of the attribute of interest in the field of scientific discovery. We first propose using attribute weights and attribute weights information gain in SBMDT that make boosting method useful for decision trees in the field of science discovery. The result of experiments on microporous aluminophosphates database about inorganic chemistry is exciting. A new template diethanolamine was found to synthesize aluminophosphates with 12-membrane ring channel structure for the first time in the world.