Predicting breast cancer using ant colony optimisation / Siti Sarah Aqilah Che Ani

Breast cancer is one of the common reasons of death for women in every country, including Malaysia, and the number of breast cancer cases is rising every year worldwide. This type of cancer happens when there exist lumps or additional tissue mass in the breast which are also known as tumors. The cancer cells can be dangerous if it is a benign cell but not a cancerous cell if it is a malignant cell. In order to classify the breast cancer cells, an accurate and effective classification model is needed. Hence, the main purpose of this study is to develop a model of classification for breast cancer cell prediction. This study implements a machine learning algorithm called Ant Colony Optimization (ACO) algorithm to develop an accurate classification model for predicting breast cancer cells. In this study, the ACO algorithm will be compared with another machine learning algorithm - the J48 algorithm - to compare which brings more precise and effective results. Besides, in this study, the Ant-Miner system also plays an important role, which can train the data several times to achieve the highest percentage of predictive accuracy and make comparisons to achieve a good classification model. The results of this study have shown that a classification model using ACO is as comprehensible and accurate as one using the J48 algorithm due to the predictive accuracy produced. Since the lowest rules number and condition numbers is produced by the Ant Colony Optimization, this study concludes that it is the most suitable algorithm in developing the classification model for predicting breast cancer cells.