Implementasi Algoritma Klasifikasi K-Nearest Neighbors untuk Prediksi Risiko Depresi pada Pengguna Media Sosial

Abstract

This study focuses on  developing  a classification model to predict the level of depression risk among social media users using machine learning techniques. This research is based on the increasing use of social media, which may have negative impacts on mental health, particularly depression, thus highlighting the importance of early detection. This study applies a quantitative approach using the K-Nearest Neighbor (KNN) algorithm as the classification method. The dataset used is secondary data obtained from Kaggle, consisting of 481 respondents, including variables related to social media usage behavior and mental health indicators. This research stages include data preprocessing, splitting the dataset into training and testing sets with an 80:20 ratio, model training, and evaluation using accuracy and F1-score metrics. The results show that  KNN algorithm is capable of classifying depression risk into several categories, with an accuracy of 38%. This indicates that the model demonstrates an initial classification capability, although its performance still needs improvement. Therefore, further development is required to enhance the model’s accuracy in supporting early detection of depression risk based on social media behavioral data