Predictive Modeling for COVID-19 Detection: A Logistic Regression Approach with Binary Data Analysis

In the quest for effective COVID-19 detection methods, this work presents an examination into the application of logistic regression on a binary dataset, containing replies to several health-related queries. Leveraging a dataset of 4309 rows and 19 columns, this research aims to construct a strong prediction model for COVID-19 diagnosis. The dataset comprises responses to questions pertaining to symptoms, travel history, and probable exposure, all recorded as binary variables (yes/no). Our findings demonstrate a promising model performance, with an amazing ROC-AUC value of 93.62%, indicating the model's great discriminative capacity. Additionally, 3.02% MSE value indicates the accuracy in predicting target values. The R-squared (R2) score of 80.78% emphasizes the potential to specify the target variable variance. This research highlights the potential of logistic regression as a beneficial technique for COVID-19 detection when applied to binary data. The results imply that such a method can produce accurate and trustworthy predictions, giving a vital contribution to the ongoing efforts to prevent the COVID-19 epidemic.