Prediction of Tumor in Mammogram Images Using Data Mining Models

Abstract

Data Mining (DM) is a process of extracting patterns from large datasets to represent knowledge, and focuses on issues related to its feasibility, usefulness, effectiveness and scalability. There are different stages for data preprocessing, namely, (i) Data cleaning (iii) Data integration (iv) Data selection (v) Data transformation, after which they are made ready for the mining task. DM techniques have contributed significantly towards several important researches and in the field of traditional sciences such as Astronomy, High Energy Physics, Biology, World Wide Web, Big Data Analytics, High Performance Computing, Cloud Computing, Medicine and some other related domains. This research work evaluates the performances of clustering algorithms by means of classifying the produced results based on the number of white color pixels, using classification algorithms. Totally, 10 attributes are taken for classification. From the results of clustering methods, the intensity of images (number of pixels) is considered as one of the significant attribute and it is included along with the existing nine other attributes. Classification techniques such as CART, SVM, Naive Bayes, JRip and J48 are used with various accuracy measures like FP rate, TP rate, Recall, Precision, ROC Area and F-measure.