Classifier Machine learning

Response surface methodologies The area under ROC curve Consequently, when classification models with imbalanced data are developed, most classifiers are that of bad ones. For imbalanced data of two categories, the category with more data is called the majority class; the minority class refers to the category with less data. Imbalanced data are common owing to erroneous decision or a rare subject, e.g., a valuable abalone in comparison with a common one. Bankrupt prediction data and credit scoring data are normally imbalanced data. As a dichoto-probability and thus fail to construct an appropriate model (Jap-kowicz & Stephen, 2002). The priori class probability refers to a probability in which an instance belongs to a certain class under general circumstances. As long as data are accumulated properly, the probability that the accumulated data belong to a certain class can be treated as a priori class probability. Consider a credit scoring data set, in which bad credit data comprises 1/10 of the entire data set; in addition, the priori class probability of bad credit and good credit are 0.1 and 0.9, respectively. Classifiers such as artificial neuron network (ANN) or decision tree (DT) construct models are based on properties from instances to identify classes. Most ⇑ Corresponding author