dataset for Text Mining-Based Empirical Analysis of How Personality Traits Affect Stock Investment

Based on the Big Five personality theoretical framework, a two-layer textual feature mapping system was constructed, firstly, the surface semantic matching was performed by the Dirichlet Distribution (LDA) topic model, and 20 potential topics were extracted (Perplexity Perplexity=198.76), and their cosine similarity with the CFPS dimensions was computed, and then, the Probit two-stage prediction model was established to compare the textual features with the CFPS scales in terms of their explanatory power for stock holding choices. After the above steps, the results of validation of personality trait indicators were obtained, as shown in Table 1.