부산대학교 도서관

Mental illness has been one of the leading factors of death by people between 15 – 29 years old. People with mental illness could prevent themselves from getting help because of the fear of what people might think of them. It is essential to raise awareness of mental illness, especially among tech workers, as one of the most common and advanced jobs. In this research, the dataset from the Open Sourcing Mental Illness, LTD which began in 2016 was analyzed using data science techniques and using the Support Vector Machine (SVM) and Fully Connected Neural Networks (FCNN). The two algorithms were then compared to find the most accurate model to predict the risk of mental illness, especially among IT workers. This research found that for 864 data train and 371 data tests a final accuracy rate of 77.089% for the SVM and 76.011% for the FCNN was obtained. The dataset used has a target prediction that consists of data with a categorical data type, so it is categorized as multiclass classification. The experimental results in this research show that on predicting the mental health of tech worker from 1235 sample data, the machine learning that represented by SVM shown a better solution effect.