Applying Random Forest Algorithm to Predicting the Stock Price Trend of IC Design Companies

IC design companies play an important role in Taiwan's semiconductor industry and provide a critical pillar for global semiconductor development. This study takes IC design stocks as the research target, and proposes a stock price prediction method based on the random forest (RF) algorithm. The RF algorithm classifier trains the prediction model through parameter adjustment, and then through feature screening to improve the accuracy rate. It generates the final result of predicting the stock price. The experimental data collected closing price data for input materials and technical indicators as feature values. The data is obtained from XQ Global Winner, and the research period is daily data from 2015/01/05 to 2022/12/30. The results found that after screening the model features, the stock price prediction trend can significantly improve the accuracy and AUC, and improve the model's explanatory power.