Applications of DQN in the Correlation Analysis between Venture Capital and Residents' Happiness

Abstract

This paper introduces the deep Q-network (DQN) algorithm into the field of correlation analysis between venture capital and residents' happiness, and analyzes the dynamic impact mechanism of capital allocation behavior on people's happiness by constructing a reinforcement learning framework of ''state observation-investment decision-environmental feedback". Based on the panel data of 20,260 households, the significant advantages of DQN in capturing nonlinear association are verified: its prediction accuracy, extreme risk control and model explanatory power are superior to those of the comparison algorithm. The key findings reveal that the marginal gain of stock investment on the happiness of ordinary families is 1.2 points, while bond allocation significantly improves the happiness of the elderly (+ 2.1 points); derivative investment shows an inverted U-shaped adjustment curve, and the happiness drops sharply when the risk preference coefficient exceeds 0. 85. This study provides a quantitative basis for differentiated capital allocation policy, and promotes the transformation of venture capital from economic value creator to social value creator.