Logistic Regression

This chapter delves into logistic regression, a fundamental machine learning technique for classification problems, using KNIME as the primary analytical tool. It begins with basic concepts, explaining how logistic regression differentiates from linear regression by using the Sigmoid function to predict binary outcomes, such as determining eligibility for a scholarship based on exam scores. The chapter emphasizes the creation of decision boundaries and illustrates the use of dummy variables for categorical data handling. Substantial focus is given to model evaluation techniques, including ROC (receiver operating characteristic) curves and F1 scores, enhancing the understanding of model performance. Practical implementation is anchored through a step-by-step guide to building a logistic regression model in KNIME, covering aspects from data reading and preprocessing to model training, testing, and evaluation. A significant part of the chapter is dedicated to a case study: predicting survival on the Titanic using logistic regression. This real-world application provides a comprehensive understanding of the logistic regression process, from handling data types and missing values to interpreting model coefficients. The chapter concludes with discussions and practice exercises, consolidating the concepts learned and their application in logistic regression analysis.