Statistical Analysis of Social Media on Mental Health in Young Adults

Abstract

This research explores the correlation between the behavior of social media use and emotional well-being in young adults in a quantitative correlational study.The data is in the form of self-reported indicators, such as platform preference, frequency of engagement, the frequency of activities within the platform, and the attitude of emotional states, which are Happiness, Anxiety, Anger, Sadness, Boredom, and Neutral.. Python libraries were used to process the data and analyze it with the help of exploratory data analysis (EDA) to guarantee that the information is credible and shows trends in behavioral-emotional changes. The key variables were on which feature extraction concentrated. Type of platform, time spent using it on a daily basis, and metrics of user engagement. The machine learning classifiers used were five: the Logistic Regression, the Random Forest, the XGBoost, the Support Vector Machine (SVM), and the K-Nearest Neighbors (KNN) classifiers which were used to predict the emotional state using behavioral features.The outcomes of the experiments proved that ensemble learning models were far more successful than random and instance-based models in terms of the accuracy of XGBoost (88% accuracy) and Random Forest (85% accuracy). The EDA showed that Happiness and Neutral emotions were associated with having more social interaction, but Anxiety and Sadness had more association with longer screen time and reduced levels of engagement. Such results demonstrate the possibilities of machine learning to determine emotional inclination in the form of social media behavior. The conclusion of the study is that predictive analytics can be essential in the promotion of digital health and the development of the intervention plan designed to engage in social media in a healthier way.