Binary northern goshawk optimization for feature selection on micro array cancer datasets

Northern Goshawk Optimization (NGO) is a recently proposed swarm-based optimization method that hunts like a northern goshawk. However, while the approach excels with many benchmark functions, it is incapable of dealing with the binary optimization problem. We proposed a binary variation of NGO for feature selection (FS) issues in classification tasks. We employed S and V-shaped transfer functions (TF) to convert continuous data to binary values. The suggested model is evaluated based on six high dimensional micro array cancer datasets using the benchmark evaluation measures such as accuracy, fitness and number of features selected. To demonstrate the effectiveness of the suggested model, it is compared to traditional and recent binary versions metaheuristic algorithms. According to the findings, the S-shaped transfer function surpasses other transfer functions and classical models.