Human dynamics in repurchase behavior based on comments mining

Abstract

Hundreds of thousands of individual deals and comments are analyzed to ask: what kinds of patterns appear in their repurchase process? Our results suggest that, in the empirical description, the intervals between two consecutive purchases obey a power-law distribution. Notwithstanding a wide range of individual preferences, shoppers’ repurchase behaviors show some similar patterns, called long-scale quiet and short-scale emergence, and the alternating appearance of them form an endless chain in repurchase. In agreement with the empirical results, these short-scale and long-scale patterns suggest an adaptive model with alterable exponents complying with a power-law distribution. And it also implies that each user behaves his own intrinsic pattern such as unique repurchase intensity and silence-emergence cycle, which contributes to customer life-time value from the new view of dynamics and repurchase cycles.