This paper proposes a quantitative comparison between two methods of Machine Learning: Prototype Adjusting Method and Cluster Ensemble. The partitions were obtained by the fuzzy c-Means Algorithm (FCM) applied to five literature datasets. The results showed that the Prototype Adjusting Method is promising; it achieved better solutions than Cluster Ensemble, as well as, a gain over than 15% in the final partition in the respective ensemble in 60% of databases.