A Comparison Study of Data Mining Algorithms for blood Cancer Prediction

Cancer is a common disease that threats the life of one of every three people. This dangerous disease urgently requires early detection and diagnosis. The recent progress in data mining methods, such as classification, has proven the need for machine learning algorithms to apply to large datasets. This paper mainly aims to utilise data mining techniques to classify cancer data sets into blood cancer and non-blood cancer based on pre-defined information and post-defined information obtained after blood tests and CT scan tests. This research conducted using the WEKA data mining tool with 10-fold cross-validation to evaluate and compare different classification algorithms, extract meaningful information from the dataset and accurately identify the most suitable and predictive model. This paper depicted that the most suitable classifier with the best ability to predict the cancerous dataset is Multilayer perceptron with an accuracy of 99.3967%.