Using Candlestick Charts to Predict Adolescent Stress Trend on Micro-blog

Abstract Surveys conducted by psychological institutions show that many teenagers suffer from stress in study, affection, and appearance shape nowadays. Coping with stress has become a major concern for teenagers. With the development of social media, the huge amounts of teens’ tweets on micro-blog make it possible to sense teens’ stress and predict their stress level change through their tweeting contents and behaviors. The finding results enable teachers and parents to know teenager’ stress situations better, be aware of the possible stress level change in advance, and take actions to help stressful teenagers relieve stress. In this paper, we propose to use the candlestick charts which are widely adopted in stock analysis to predict teens’ stress level change on a micro-blog platform. As different individuals have different stress changing characteristics, we define a stress pattern as a sequence of candlestick charts, and do stress pattern matching to judge if a candlestick chart with reversal signals will result in the trend reversal. Our experimental results verify the effectiveness of our candlestick charts based method in predicting teens’ stress level change, compared to time series, MACD (Moving Average Convergence/Divergence), and KDJ (Stochastic Oscillator) based prediction approaches.