Using Logistic Regression Approach to Predicating Breast Cancer DATASET

Abstract

The aim of this study is to predicate breast cancer by using three approaches: Multilayer perceptron, multiple linear regression, and logistic regression, then make a comparison between the performance of these approaches to decide the best approach to analyses the Wisconsin breast cancer dataset. In this study, we used the Wisconsin breast cancer dataset (699 instances and 11 attributes). Three approaches (multilayer perceptron, multiple linear regression, and logistic regression) were applied to the dataset to predicate the Wisconsin breast cancer. Performance was calculated for each approach and a comparison was made between the three approaches to see which is better. From the performance comparison between multi-layer perceptron, multiple linear regression, and logistic regression, the Wisconsin breast cancer dataset is best to analyses using multi-layer perceptron, with 100% compared to logistic regression 97.6% and multiple linear regression 84.4%. The computerized and especially logistic regression model could be useful to the predicate for many fields of science. For this reason, it’s used abundantly in the medical field specially to build a model to predicate breast cancer. The conclusion from this experiment showed that not all the independent variables have the coefficient effect with the dependent variable. However, the logistic regression approach is the best approach to analyses the Wisconsin breast cancer dataset rather than the multi-layer perceptron and multiple linear regression approach.