Hybrid Approach using Extreme Gradient Boosting (XGBoost) and Evolutionary Algorithm for Cancer Classification
An Evolutionary Algorithm named as Group Counseling Optimizer (GCO) in tandem with Extreme gradient boost (XGBoost) to classify cancer in microarray data is employed.
Abstract
Cancer is a leading cause of mortality globally. World Health Organization estimates that around 10 million people die from cancer in 2020. Cancers of the colon, breast, lung, central nervous system, cervix, and prostate are quite common. In most cases, cancer may be efficiently treated if caught early. The biological data included in microarrays is very informative. This data analysis aids in the identification and treatment of difficult illnesses. It is quite difficult and time consuming to train a model with such a large number of features, when they are all input directly. Therefore, we employ an Evolutionary Algorithm named as Group Counseling Optimizer (GCO) in tandem with Extreme gradient boost (XGBoost) to classify cancer in microarray data. At the outset, we have the Extreme Gradient Boosting ensemble-selected features. In this step, irrelevant features are eliminated, and a set of optimal characteristics for detecting cancer has been generated. In a second usage, the evolutionary method is used for cancer classification.