The performance of the classification models are highly degraded on large dataset with high dimension. The high dimensional dataset has both relevant as well as irrelevant features results in performance degradation of the classification model. Moreover, more number of datasets is imbalanced in nature. The imbalanced data also poses hindrance for the classification models. The imbalanced data set leads classification performance bias towards the majority class. In this work, PCA with smote is applied to derive an effective subset of lung dataset. The PCA reduces the dimensionality of the data set into lower dimension. The SMOTE is the technique that creates synthetic samples on the dataset. PCA eliminates the irrelevant features from the dataset and SMOTE creates new synthetic samples to increase the number of representative samples in minority class. Finally, SVM classifier is applied on the pre-processed dataset as well as performance of the model is compared using evaluation metrics. The experimental results proved the effectives of the proposed methodology in terms of accuracy, precision, recall, false positive rate