回归分析已成为水土保持研究重要的数学处理手段。但在水土保持回归研究中 ,存在多元回归偏少、不重视回归检验、数学概念理解不清等问题 ,提出在研究过程中 ,当量纲不一的情况下 ,应该用回归系数进行比较。在非线性回归中 ,应列出相关指数 ,而不是相关系数。对试验中未预见到的干扰情况或正常试验出现的异常值 ,做好重点记录 ,并对此测定数据进行重点处理 ,分析其出现的原因 ,判断其属性 ,或对其消除 ,或对其进行重复测定补充验证 ,加大样本量后 ,再决定取舍。当试验完成后 ,对数据进行回归分析时 ,要进行检验 ,若检验通过 ,方可记录于技术报告。否则要分析原因 ,从另一角度进行阐释 ,有待于进一步研究 ,有助于研究的正确发展以及新问题的发现 Regression analysis has become an important mathematical treatment of soil and water conservation. However, in the study of regression of soil and water conservation, there are many problems such as multiple regressions, neglect of regression test and unclear understanding of mathematical concepts. It is suggested that regression should be compared in the process of research when the dimensions are different. In non-linear regression, the relevant index should be listed instead of the correlation coefficient. Make an important record of unforeseen disturbances or abnormal values ​​that appear in normal tests in the experiment, and focus on the measured data, analyze the reasons for the occurrence, judge its properties, eliminate them, or carry out them Repeated determination of additional validation, increase the sample size, and then decide to choose. When the test is completed, the data regression analysis, to be tested, if the test passed before recorded in the technical report. Otherwise, it is necessary to analyze the reasons and explain from another perspective, which needs further study and is conducive to the correct development of research and the discovery of new problems