相关性分析、聚类分析、因子分析的可靠性对比——以嘎拉勒和躬琼左波两条剖面为例

Reliability Comparison of Correlation,Clustering and Factor Analyses for Lithogeochemistry:Examples of Two Profiles in Galale and Gongqiongzuobo,Tibet

文章简单介绍了数学分析方法中相关性分析、聚类分析、因子分析的算法原理,并利用该方法对嘎拉勒外围和躬琼左波两条剖面的岩石地球化学数据进行了处理。对样品进行了R型聚类分析和R型因子分析,并对样本进行了Q型聚类分析和Q型因子分析。所测结果结合矿物学相关知识对矿床成因进行分析,并对分析结果与前人对该矿床的认识进行拟合。从而得出R型因子分析和聚类分析对矿床成因可做某种程度的解释,Q型因子分析和聚类分析对深部岩体预测具有部分解释作用。

The article simply introduces the mathematical analysis methods of correlation,cluster and factor analyses,which are used to process geochemical data of rocks on two profiles in Galale and Gongqiongzuobo,Tibet. We make analyses of R-type clustering and factor,and Q-type clustering and factor on the samples from the profiles. Combined with the mineralogy,we use these analyses to explain origin of ore deposits and fit for the understanding of previous work. It is concluded that R-type factor and clustering analyses permit to interpret the genesis of ore deposits to some extent,and Q-type factor and clustering analyses are able to predict deep magmatic rock bodies with certain reliability.