Data mining is dedicated to retrieve knowledge from massive datasets by utilizing automated algorithms. However, due to the characteristic of automation processes, current data mining approaches can hardly allow the user to visually understand, explore and optimize the datasets and the computation process. Recently an increasing number of researchers in the field of visualization have been focusing on visualization-based interactive data mining approaches. With the assistance of visualization, users can gain insight and perform exploration from datasets and the mining results intuitively. In this paper we compare data mining and visualization process from the aspect of knowledge discovery. Additionally we classify the recent works into two main categories: 1) visual-enhanced general data mining approaches, and 2) application-based approaches. Additionally we propose a set of future challenges according to recent related conferences.