Predicting the interference of mental health illness at work productivity in the technology industry using machine learning methods

Abstract

Advancements in Machine Learning technologies are progressively providing enhanced benefits, especially in the domain of making accurate predictions applicable to diverse scenarios. One such scenario of critical interest is mental health and its impact on workplace productivity, particularly within the technological industry. This thesis aims to predict work interference arising from mental health issues in the technology sector using Machine Learning techniques. Seven carefully selected Machine Learning classification models were applied to a dataset sourced from the non-governmental organization known as Open Sourcing Mental Illness. The dataset underwent strategic cleaning and preparation, ensuring the retention of all relevant variables necessary to meet the requirements of each model and facilitate successful deployment. Subsequently, the models were rigorously evaluated using various measures of prediction, including Accuracy and Precision, among others. The research identified the 'Gradient Boosting Classifier' as the most effective model, exhibiting superior performance across the majority of prediction measures, including an 83.2% accuracy. This investigation also uncovered similar limitations to those observed in prior machine learning studies related to mental health, as discussed in the Literature Review. However, the findings contribute valuable insights into the application of Machine Learning in predicting work interference due to mental health illness, particularly within the dynamic landscape of the technology industry.

Date of Award	24 Jun 2024
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Pedro Afonso Fernandes (Supervisor)