Data-Driven Exploration of Pandemic’s Psychological Impact and Lifestyle Changes Through Clustering Approach

The COVID-19 pandemic has profoundly impacted mental well-being and lifestyle behaviours. This paper employs data-driven methodologies and clustering algorithms to analyze the COVID-19 psychological impact and lifestyle changes. Main focus is on the intricate psychological dynamics and stratification of individuals based on shared lifestyle adaptations. Beginning with meticulous data preparation, including cleaning and transformation, we utilize feature engineering to identify individuals with notable lifestyle transformations. Supervised learning techniques, such as Decision Trees and Random Forests, are applied for prediction and categorization of individuals experiencing substantial changes. Unsupervised learning, involving K-Means and DBSCAN algorithms, reveals latent intricacies and creates clusters representing shared lifestyle shifts. The Decision Tree Classifier exhibits an outstanding accuracy of 99.15%, surpassing both Logistic Regression (82.13%) and Random Forest (98.72%) models. DBSCAN demonstrates a notably higher Silhouette Score of 0.530, indicating well-separated clusters with distinct boundaries, outperforming K-Means (0.277) and Hierarchical Clustering (0.278). The insights contribute to understanding the psychological reverberations of the pandemic and adaptations in lifestyle dynamics, aiming to provide meaningful insights for mental well-being in the current and future challenges.