A hybrid recommendation model for successful R&D collaboration: Mixing machine learning and discriminant analysis

Seeking to stimulate and improve the rate of success of R&D collaboration by SMEs, this study developed a method of recommending types of external collaboration organizations that are optimal partners for SMEs. We began by examining the current data on R&D collaboration by partner type to effectively classify the types of R&D partners engaged with South Korean SMEs. Next, we applied machine learning and discriminant analysis to develop a hybrid model for recommending firms that will likely achieve high satisfaction from collaboration with four representative types of R&D partners (universities, public research institutes, large firms, and SMEs). Lastly, we used new data that had not been included in the model development stage, to perform additional evaluations of the model. In our research results, the hybrid recommendation model, designed to identify SMEs that will achieve high satisfaction by R&D partner type, demonstrated outstanding accuracy exceeding 91%. By applying the model proposed in this paper, firms will be able to select their R&D partner types more efficiently and improve the likelihood of achieving success in R&D collaboration. Meanwhile, those responsible for implementing public policies may use the proposed model to improve the efficiency of public investments that support R&D collaboration.