Clustering Mental Health on Instagram Users Using K-Means Algorithm

Abstract

The use of Instagram too often can have an impact on the mental health of its users. Mental health that is not good requires early treatment so that it does not have a widespread impact on other health. Mental illness requires a professional to treat it as an effort to prevent a disease from getting worse. However, the stigma attached to sufferers is one of the significant causes behind the reluctance to seek treatment. Therefore we need a way so that Instagram users can find out for themselves the condition of their mental health. One way is to do Clustering the use of Instagram so that it can provide an early indication of a person's mental health. From the proposed model we can find out the categories of 600 respondents who were collected using a questionnaire with 10 main attributes. The proposed model is k-means with 3 clusters determined using the elbow method. In this study, the last centroid obtained through calculations was used to evaluate the k-means by comparing the results of the k-means calculations with the results of psychologists. The results of the K-means evaluation have an accuracy of 73.83% so that the last centroid can be applied to web-based applications that have been created. This mental health clustering model is expected to be able to help the community to get mental health conditions early and reduce the negative stigma that exists and can be used as evaluation material in using social media more wisely.