Abstract :
In coal analysis, there exists high dimensional feature information in the dataset. However, there are several irrelative or redundant features. Feature selection algorithm could select the relative attributes, but it is possible that some important features are deleted. In this paper, we propose hierarchical clustering based logistic regression algorithm. It firstly selects multiple groups of features using random selection, and builds multiple models. After model built, it clusters the models and makes the similar models into one clustering. Finally, we use the logistic regression clustering to complete the forecast work for future dataset.
