1. SAS INSIGHT启动:
方法1:Solution→Analysis→Interactive Date Analysis
方法2:在命令栏内输入insight
方法3:程序编辑窗口输入以下代码,然后单击 Submit按钮;
Proc insight;
Run;
用 sas insight做直方图、盒形图、马赛克图。
直方图:Analysis→Histogram/Bar Chart
盒形图:Analysis→Box plot
马赛克图:Analysis→Box plot/Mosaic plot(Y)
散点图:Analysis→Scattery plot(Y X)
曲线图:Analysis→Line plot( Y X)
旋转图:Analysis→Rotationg Plot
曲面图:Analysis→Rotationg Plot 设置 Fit Surface
等高线图:Analysis→Countor plot
包括:直方图、盒形图、各阶矩、分位数表,直方图拟合密度曲线,对特定分布进行检验。
第一部分为盒形图,第二部分为直方图,第三部分为各阶矩,第四部分为分位数表。
A:参数估计:给出各种已知分布(正态,指数等),只需要对其中参数进行估计;
Curves→Parametric Density
B:核估计:对密度函数没有做假设,曲线性状完全依赖于数据;
Curves→Kernel Density
Curves→CDF confidence band
Curves→Test for Distribution
Analysis→Fit(Y X):分析两个变量之间的关系
Analysis→Fit(Y X)
Analysis→Fit(Y X)
Analysis→Multivariate
Analysis→Multivariate
方法1:Solution→Analysis→Analyst
方法2:在命令栏内输入analyst
Statistics →Descriptive→Summart Statistics 只计算简单统计量
Statistics →Descriptive→Distribution 可计算一个变量的分布信息
Statistics →Descriptive→Correlations可计算变量之间的相关关系
Statistics →Descriptive→Frequency counts 可计算频数
Statistics →Table Analysis
Statistics →Hypothesis tests →One-Sample Z-test for a mean
推断该样本来自的总体均数μ与已知的某一总体均属μ0是否相等
Statistics →Hypothesis tests → One-Sample t-test for a mean
Statistics →Hypothesis tests →One-Sample test for a proportion
Statistics →Hypothesis tests→One-Sample test for a variance
Statistics →Hypothesis tests →Two-Sample t-test for means
Statistics →Hypothesis tests →Two-Sample paired t-test for means
Statistics →Hypothesis tests →Two-Sample test for proportions
Statistics →Hypothesis tests→Two Sample test for variance
Statistics →ANOVA→One-Way Anova
Statistics →ANOVA→nonparameter one-way Anova test
Wilcoxon法、Median法、Van der Waerden法、Savage法。
Statistics →ANOVA→Factorial Anova
Statistics →ANOVA→Linear Model
Statistics →Regression→simple
Statistics →Regression→linear
Statistics →Regression→logistic
Proc print data = sasuser.score; //数据库.数据集 Run; |
Proc print data = sasuser.score; Var name math Chinese; //变量 Run; |
Proc print data = sasuser.score noobs; //去掉第一列(观测序号) Var name math Chinese; Run; |
Proc print data= sasuser.score; Where sex in(‘f’); //通过where语句 Run; |
Proc print data = sasuser.score noobs label; Title ‘女生成绩单’; Label name =‘姓名’ Sex =‘性别’ Math = ‘数学’ Chinese = ‘语文’ English = ‘英语’; Where sex in(‘f’); Run; |
Title “the sas system”; //恢复系统标题 |
Proc print data = sasuser.score; Footnote = ‘分数列表’; //加分数列表的脚注 Run; |
Proc sort data = sasuser.score; By sex; Run; Proc print data = sasuser.score; //使用by分组输出前用sort排序 By sex; Run; Proc print data = sasuser.score; Sum math; Run; |
Proc tabulate data =数据集名称; Class 分类变量; Var 分析变量; Table 页面说明 行维说明 列维说明/选项; Run; |
Proc sort data = 数据集名称; //默认升序排列 By 变量名; Run; |
Proc sort data = 数据集名称; By descending 变量名; //降序排列 Run; |
Proc means data = sasuser.stock; Var price; Run; |
Proc univariate data =数据集; Var 分析变量; Run; 结果: Moments:统计量的各阶矩,例如一阶矩就是均值,二阶矩就是方差等; Basic Statistical Measures:基本统计量; Tests for location:检验均值是否为零; Quantiles:分位数表; Extreme Observations:极端观测值。 |
Proc freq data =数据集名; Tables 变量名; Run; 结果: 变量取值、频数、百分比、累计频数、;累计百分比 |
Proc corr data =数据集; Var 变量名 变量名; Run; 结果: 简单统计量 相关系数及p值 |
Proc gplot data = 数据集名称; Symbol 曲线类型; Plot 竖轴变量*横轴变量; Run;
Proc gplot data = sasuser.score; Symbol I = none v=star; Plot English*Chinese; Run;
|
Proc gchart data = 数据集名称; Vbar/pie/block =变量; Run; |
Proc g3d data =数据集; Plot 变量x*变量y=变量z; Run; |
Proc gcontour data =数据集名; Plot x*y=z; Run; |
Proc univariate data= sasuser.stock normal; Var eps; Run; |
Proc univariate data= sasuser.stock normal; Var eps; Histogram eps; //画出直方图 Probplot eps; //画出概率分布图 Run; |
4.2.1如果一个变量服从正态分布,那么可以用t检验来对变量进行均值检验
Proc ttest data =数据集 ho = 均值; Var 检验变量; Run; |
4.2.2t检验还可以检验方差相同的两个独立样本均值是否相等
Proc ttest data =数据集; Class 分类变量; Var 检验变量; Run; 结果 第一部分简单统计量 第二部分t检验结果 第三部分两者方差是否相等检验 |
T检验要求两个独立样本都必须服从正态分布,如果不服从正态分布,则无法进行t检验。这时可用非参数的方法,常用的非参数方法是NPAR1WAY过程,它是 noparameter 1 way缩写。
4.4.1 REG过程
Proc reg data = 输入数据集 选项; Var 变量列表; Model 因变量 = 自变量列表; Print 输出结果; Plot 诊断图形; Run; |
指明模型的表达式并给定系数初值
4.5.1单因素方差分析
Proc anova data =数据集名称; Class 因素; Model 实验结果 =因素; Run; |
Proc anova data =数据集名称; Class 因素; Model 实验结果 =因素; Means brand; Run; |
Proc anova data =数据集名称; Class 因素; Model 实验结果 =因素; Means brand/t; //t检验 Run; |
Proc anova data =数据集名称; Class 因素; Model 实验结果 =因素; Means brand/bon; //bonferroni t检验 控制第一类错误的概率,但是具有较大第二类错误概率 Run; |
Proc anova data =数据集名称; Class 因素; Model 实验结果 =因素; Means brand/regwq; //regwq检验 控制第一类错误的概率 Run; |
Proc anova data =数据集名称; Class 因素; Model 实验结果 =因素; Means brand/tukey; //tukey检验 控制第一类错误的概率,但是第二类错误概率通常高于regwq检验 Run; |
4.5.2多因素方差分析
4.5.3列联表检验
Proc freq data = 数据集; Tables 因素a*因素b / chisq; Weight 实验结果; Run; |
因变量—Depender (Y)
自变量—Independent (X1 X2…)
数据分析咨询请扫描二维码