Data mining techniques are being used for breast cancer classification and good performance accuracy has been obtained while using the techniques individually or as ensembles. A notable problem is the skewed nature of the data which leads to imbalance among the output classes. The minority class being the negative class usually are smaller in number than the positive majority class. This usually leads to a moderate accuracy value for the classifier. The correct classification of minority cases is a significant problem. All classes need to be given equal importance during classification. In this study a hybrid model based on Logistic Regression is implemented with class balancing and ant search techniques and the performance is evaluated on the two class Wisconsin breast cancer dataset. A performance accuracy of 99.4% was obtained.