The sheer volume of user behavioral data captured on online social media provide credit agencies with unprecedented opportunities to tap into the low-cost social intelligence for conducting online credit scoring. However, traditional machine learning techniques simply cannot scale up with the large-scale social media data. The main contribution of the work reported in this paper is the development of a novel large-scale data analytics methodology that leverages readily available social media data for enhancing online credit scoring. In particular, the proposed methodology is underpinned by a parallel topic modeling method for user behavioral pattern mining. Based on real-world data crawled from Sina Weibo, our experimental results show that the proposed large-scale data analytics methodology can effectively and efficiently analyze user behavior patterns from online social media. Moreover, it outperforms traditional credit scoring methods.