Research on Financial Data Mining and Investment Decision Optimization under the Background of Digital Transformation

Abstract

Under the background of digital transformation, traditional investment decision-making methods are facing challenges due to insufficient information processing ability and time lag. Based on the ecological transformation of financial data, this study constructs an analytical framework integrating data collection, data mining and investment decision optimization. By comparing the performance of three algorithms, XGBoost, Long Short Term Memory (LSTM) and random forest, it is found that XGBoost is significantly superior to other models in terms of mean absolute error (MAE), root mean square error (RMSE) and other indicators, and its gradient lifting framework can effectively capture the nonlinear relationship between high-dimensional features. Based on the signal generated by XGBoost, a two-stage investment strategy is constructed: in the first stage, the income forecast signal is generated by comprehensive feature engineering, and in the second stage, the asset weight is allocated by means of mean-variance optimization model. The empirical results show that this strategy achieves an annualized rate of return of 20.56% and a Sharp ratio of 0.81% in the out-of-sample interval, and the maximum retracement is controlled to -20.36%, which is superior to the benchmark index and the single model strategy. The research verifies the effectiveness of multi-source data mining technology in improving the scientific nature of investment decision-making and risk adjustment income, and provides an intelligent decision-making tool for financial institutions and individual investors.