Issues related to mental health are increasingly becoming legitimate medical concerns in the workplace. To mitigate mental health concerns among employees, companies need to identify the points of view that bear most of the blame. Some preprocessing is performed on an open-source survey data set before selecting features for language processing. Techniques such as logistic regression, Naive Bayes, and random forest are used for classifications to determine whether an employee needs treatment for mental health issues or not. The activities were modeled as a bag of words using a probabilistic Naive Bayes (NB) classifier on training data with an accuracy of 92.7% and a random forest with an accuracy of 99.7% to predict mood outcomes and logistic regression with 99. 8% accuracy. The analysis is divided into two parts: a personalized model and a generic model in which we combined the data from all participants. Finally, a voting classifier is used to add all of the findings from every classifier that is fed into it and forecast the result based on the most notable majority of votes. Combining the aforementioned techniques, the voting classifier improves our accuracy to 90.48%.