Analysis of Principal Component Analysis Algorithm for Various Datasets

Principal component analysis (PCA) has been used successfully as a multivariate statistical process control (MSPC) tool for detecting faults in processes with predominantly found variables. However, in machine learning, predicting the result or object hugely depends on the proper dataset. Generally, the available dataset is bulk and consists of redundant information. To make any prediction, we need to have a mechanism to remove unwanted information from the dataset for high accuracy. If we use the statistical method, there is a high chance of enormous data loss while processing. Hence in this paper, we use PCA to reduce the high dimensional data set to a smaller number of modes or the structure. We implemented PCA for a simple 2 X 2 process using python for datasets such as breast cancer, wine data set, Digits, and Iris dataset to understand the impact of PCA on various application datasets. Our results show that the monitoring performance of PCA is vastly better, with 98.15% accuracy for the Wine data set, 91.0S% for breast cancer, 75.90% for digits, and 96.90% for the Iris dataset.