En-InfoFi: An ensemble classifier based on Information Gain and Firefly for Cancer classification

Cancer is now becoming a prominent factor in increasing the death toll. As per WHO the death toll due to Cancer is approximately 10 million which is about 38% of total death. This can be avoided by early diagnosing the disease at its early stage and machine learning is nowaday playing an important role in doing so. The basic need for machine learning is the genetic dataset or microarray data. The primary characteristic of microarray data is the high dimensionality which also becomes a primary concern while implementing the machine learning algorithms. Applying the machine learning algorithms causes many issues out of which the small sample size is one of the biggest ones that can decrease the performance of the model. To sort this issue the feature selection algorithms are introduced. Though the feature selection algorithms reduce the dataset dimensionality, still some unnecessary features remain. For optimal feature selection, the metaheuristic and nature-inspired optimized algorithms are implemented in addition to the feature selection technique which results in a hybrid model. But in the case of the hybrid model, only one classification method is considered. So, to improve the model performance ensemble machine learning models are introduced which integrate multiple classification approaches to a single problem space. In this current research work, ensemble machine learning based on Support Vector Machine (SVM), Random Forest (RF), and Naive Bayes (NB) are considered the primary classifiers along with the Information Gain feature selection and Firefly Algorithm optimization algorithms. Based on the individual performance in terms of accuracy the SVM, RF, and NB are considered to form an ensemble model. The Proposed En-InfoFi is evaluated on 3 different types of cancer datasets including Breast Cancer, Lung Cancer, and ALL-AML cancer datasets.