Machine Learning Aided Fused Feature Selection based Classification Framework for Diagnosing Cervical Cancer

A dataset with huge number of attributes, redundant and irrelevant features could be challenging for gaining acceptable accuracy in Machine Learning [ML] based classification approaches. Medical datasets like cervical cancer data likely to have high dimensional, redundant and irrelevant attributes with mislaid values and some of them incline to have imbalanced target classes. Efficient feature selection techniques could resolve these issues and advance the performance of the classification models with advanced precision in classification by accomplishing computational efficiencies. The execution of effectual feature selection methods on cervical cancer datasets could support to attain a reduced and enhanced feature subset. Besides, the competences of filter and wrapper approaches can be integrated to acquire a finest and effectual feature subset for classification progression. This research is intended for diagnosing cervical cancer and the dataset used in this work is having missing values, redundant features and imbalanced target classes. Hence this research purposes to handle these issues through integrated feature selection approach to attain an optimal feature subset. The subsets attained through this fused approach can be employed in augmented prediction process. The best and an optimal feature subset can be decided based on the performance efficiency of the classifiers in predicting the results. This fused approach is essential in bio medical and bio informatics datasets like cervical cancer datasets where data classification with progressive accuracy is more challenging. Consequently, the aim of this research is to propose a comprehensive framework with fused feature selection process to accomplish optimal feature subset with precision in classification and to provide computational efficiencies for cervical cancer diagnosis.