Performance of Lung Carcinoma in Classification Neural Network with Pre Processing Using WEGA

Abstract

Data pre processing ease the mining procedure by removing the insignificant information and errors that may arise while entering the data manually. The data collection method is not strict so there accompanies missing and incorrect values, irrelevant variables, data with out of range etc. These have significant impact and minimize the accuracy of the mining process. Generally accuracy in the case of medical research must reach to the extent. There are many factors affect the analysis on the given task. The precise representation and quality of the dataset is vital. If there exists more irrelevant and redundant information the meaningful discovery of knowledge is a big question. Pre processing is a prominent way for the data preparation and thus it the earlier stage in mining. It includes many variant procedures according to the problem of the set. The output is taken as the direct training set for further research. This research analyse the Lung cancer dataset with fifteen attributes by applying pre processing method attribute evaluation. This method reduce the dimensionality, file size and time taken for the analysis by considering only on the most relevant variables. The work is carried in the WEKA tool as it has enormous procedures for data preparation. The performance before and after pre processing is discussed with suitable metrics.