Applications of Feature Engineering and Logistic Regression to Analyze Microchip Performance

Using key features of machine learning to calculate, generate, and enhance a nonlinear decision boundary

Abstract

Machine learning is an adaptive concept that has applications in several other industries. Businesses can use it to create real-time and predictive analyses of corporate data to identify consumer trends, calculate product success rates, visualizing short- and long-term business models, and more. This paper takes machine learning algorithms to analyze microchip performance. Beginning with an introduction of fundamental concepts of machine learning — including feature, labels, and classification vs regression models — the paper’s example is introduced with a scatterplot visualization, a definition of the loss function binary cross entropy, and the parameters of the dataset. The third and fourth sections explain the critical function of feature engineering and its utility in refining the decision boundary to best separate the data classes with minimal error. The final section reiterates the main idea that machine learning is ubiquitously applicable and, like the example in this paper, has many cross-industrial functions.