**6. Case study on variants of PCA in mobile malware detection**

To explore the dimensionality reduction using normal PCA and its variants for mobile malware detection in the CICMalDroid\_2020 dataset [11] are experimented. The dataset is taken from the University of New Brunswick (UNB), Canadian Institute of Cyber Security (CIC). The dataset consists of 11,598 records with 471 feature attributes. The dataset commences with 17,341 Android samples gathered from VirusTotal, the Contagio security blog, AMD, MalDozer, and other sources. Samples were obtained between December 2017 and December 2018. Detecting Android apps with malicious data points is crucial for cyber security specialists. There are five key

*Evaluation of Principal Component Analysis Variants to Assess Their Suitability… DOI: http://dx.doi.org/10.5772/intechopen.105418*

categories in the dataset includes, Adware, Banking malware, SMS malware, Riskware, and Benign are the different forms of malicious software. The experiment is carried out in a Python Jupyter notebook environment using sklearn library [12–16].
