The anomalies composed of ore-forming elements may not correspond with the known deposits. Some trace elements are accumulated during the mineralization process whereas some elements are depleted. The combination of the ore-forming elements and other trace elements influenced by mineralization may enhance the effect of geochemical exploration. In this study, high-dimensional data were converted into one-dimensional data for analysis by projection pursuit model based on RAGA. The projection values and the weights of elements can be calculated by this method. Moreover, important information will not be lost and the noise data will be deleted. In addition, this method is very robust. By the case study of a Pb-Zn deposit, it is found that some deposits are not in anomaly areas or some deposits do not have corresponding anomalies. However, the ranges of anomalies obtained by this method correspond with the known deposits better. The results of PPC highlights the anomalies associated with the known deposits and, moreover, the anomaly information has not been lost. The projection pursuit model combined with RAGA by using multiple elements performs better on high dimensional data and identifies stream sediment geochemical anomalies effectively and accurately.