The invention discloses an unbalanced sample data classification method. The classification method comprises the steps of inputting to-be-judged unbalanced sample data into a pre-obtained classifier model, and obtaining a classification result of the unbalanced sample data according to the output value of the classifier model, wherein the acquisition process of the classifier model is as follows:adjusting the positive penalty factor Cp of a loss function in a logistic regression algorithm by adopting a preset positive penalty factor Cip and then obtaining an optimized logistic regression algorithm; training a to-be-trained model constructed based on the optimized logistic regression algorithm by utilizing historical unbalanced sample data to obtain a classifier model. According to the invention, the positive penalty factor of the loss function in the logistic regression algorithm is adjusted, and the logistic regression algorithm is optimized. The influence of unbalanced sample data on the test result is reduced, so that the classification accuracy of the classifier is improved. Correspondingly, the invention also discloses an unbalanced sample data classification system having the above beneficial effects.