| Popis: |
Data scientists primarily seek to develop innovative methods for analysing data that are both computationally efficient and time efficient, resulting in effective data analytics. When a result, as more data is generated from different sources, the amount of data that may be analysed, manipulated, and visualised grows rapidly. We presented feature selection and evolutionary methods in this study. We also concentrate on data analytic process optimization techniques, i.e., we propose to investigate applications of nature-inspired Hybrid Genetic Algorithm (HGA) algorithms. This incorporates the GA, ACO, and PSO approaches, respectively. Feature selection optimization is a hybrid strategy that optimises chosen characteristics using feature selection techniques and evolutionary algorithms. Prior work iteratively solves this issue to arrive at an appropriate feature subset. Feature selection optimization is a domain-independent method. We employed vast dimensional data to test the suggested model. We extracted the android APK dataset from the permission API used to detect whether the applications are malicious or standard. The initial dataset contains around 351 attributes, and after the application algorithm, it reduces up to 13-16% of attributes-based essentiality and correlation of each attribute. The experimental evaluation has done on the Weka 3.7 environment for validation, and it achieves 90.70% accuracy for NB and 92.5% for SVM. |
