Mental Health and Personality Determination using Machine Learning

The purpose of this project is to use textual comments collected from users via a web application to predict their mental health and personality. Data encoding was performed using Label Encoder and Count Vectorizer. Machine learning models like Logistic Regression, SVM, Random Forest, Naive Bayes, and KNeighbors Classifier were built and models were trained on the dataset, the machine learning model which has the highest accuracy score has been selected. The predicted accuracy given by k fold Cross-Validation shows that the XGBoost classifier gives the best accuracy of 67.68 percent for personality prediction and predicts 16 distinct personality types across four axes based on their introversion, intuition, reasoning, and perceived abilities, even when tested on the same dataset. Whereas Logistic Regression predicts mental health with the maximum accuracy of 96.05 percent.