“用R进行多层回归分析”的版本间的差异
来自OBHRM百科
Lichaoping（讨论 | 贡献） （创建页面，内容为“可以用R自带的函数lm(）来做，或者用AutoModel包来完成”） |
Lichaoping（讨论 | 贡献） |
||
| 第1行： | 第1行： | ||
可以用R自带的函数lm(）来做，或者用AutoModel包来完成 | 可以用R自带的函数lm(）来做，或者用AutoModel包来完成 | ||
| + | |||
| + | ==用AutoModel包进行多层回归分析== | ||
| + | ===脚本与注释=== | ||
| + | <pre>library(AutoModel) # 启用AutoModel包。如果没有安装，请先安装。安装方法，请在R控制台输入：install.packages("AutoModel")。 | ||
| + | # 多层回归分析。因变量为y，第一层的自变量为：lag.quarterly.revenue，第二层新增的自变量为：price.index，income.level。数据来源AutoModel包自带的freeny数据 | ||
| + | run_model("y", c("lag.quarterly.revenue"), c("price.index", "income.level"),dataset=freeny) </pre> | ||
| + | ===结果=== | ||
| + | <pre>> run_model("y", c("lag.quarterly.revenue"), c("price.index", "income.level"),dataset=freeny) | ||
| + | |||
| + | |||
| + | REGRESSION OUTPUT | ||
| + | |||
| + | Durbin-Watson = 2.11 p value = 0.4729 | ||
| + | |||
| + | Partial Regression plots (all relationships should be linear): | ||
| + | |||
| + | Plot of studentized residuals: uniform distibution across predicted values requiredCorrelation Matrix for model (correlation >.70 indicates severe multicollinearity) | ||
| + | |||
| + | y lag.quarterly.revenue price.index income.level | ||
| + | y 1.0000 0.9978 -0.9895 0.9839 | ||
| + | lag.quarterly.revenue 0.9978 1.0000 -0.9894 0.9817 | ||
| + | price.index -0.9895 -0.9894 1.0000 -0.9539 | ||
| + | income.level 0.9839 0.9817 -0.9539 1.0000 | ||
| + | |||
| + | Variance inflation factor (<10 desired): | ||
| + | |||
| + | lag.quarterly.revenue price.index income.level | ||
| + | 194.85 78.58 45.52 | ||
| + | |||
| + | Standardized Residuals (observations > 3.00 problematic): | ||
| + | |||
| + | No significant outliers | ||
| + | |||
| + | Cook's distance (values >.2 problematic): | ||
| + | |||
| + | 1963.25 | ||
| + | 0.8918 | ||
| + | |||
| + | Normality of standardized model residuals: Shapiro-Wilk (p-value): 0.5586 | ||
| + | |||
| + | Model change statistics | ||
| + | |||
| + | R R^2 Adj R^2 SE Est. Delta R^2 F Change df1 df2 p Fch Sig | ||
| + | Model 1 0.9978 0.9956 0.9955 0.0212 0.9956 8360.3793 1 37 0 *** | ||
| + | Model 2 0.9988 0.9977 0.9975 0.0159 0.0021 15.4599 2 35 0 *** | ||
| + | Model 1 : y ~ lag.quarterly.revenue | ||
| + | Model 2 : y ~ lag.quarterly.revenue + price.index + income.level | ||
| + | |||
| + | Model Coefficients | ||
| + | |||
| + | Model term estimate std.error statistic p.value sig | ||
| + | Model 1 (Intercept) 0.04169 0.10138 0.4112 0.6833 | ||
| + | Model 1 lag.quarterly.revenue 0.99827 0.01092 91.4351 0.0000 *** | ||
| + | Model 2 (Intercept) 4.97077 1.24046 4.0072 0.0003 *** | ||
| + | Model 2 lag.quarterly.revenue 0.37305 0.11418 3.2673 0.0024 ** | ||
| + | Model 2 price.index -0.81887 0.17152 -4.7742 0.0000 *** | ||
| + | Model 2 income.level 0.75435 0.14454 5.2189 0.0000 ***</pre> | ||
2017年3月10日 (五) 08:23的最新版本
可以用R自带的函数lm(）来做，或者用AutoModel包来完成
用AutoModel包进行多层回归分析
脚本与注释
library(AutoModel) # 启用AutoModel包。如果没有安装，请先安装。安装方法，请在R控制台输入：install.packages("AutoModel")。
# 多层回归分析。因变量为y，第一层的自变量为：lag.quarterly.revenue，第二层新增的自变量为：price.index，income.level。数据来源AutoModel包自带的freeny数据
run_model("y", c("lag.quarterly.revenue"), c("price.index", "income.level"),dataset=freeny)
结果
> run_model("y", c("lag.quarterly.revenue"), c("price.index", "income.level"),dataset=freeny)
REGRESSION OUTPUT
Durbin-Watson = 2.11 p value = 0.4729
Partial Regression plots (all relationships should be linear):
Plot of studentized residuals: uniform distibution across predicted values requiredCorrelation Matrix for model (correlation >.70 indicates severe multicollinearity)
y lag.quarterly.revenue price.index income.level
y 1.0000 0.9978 -0.9895 0.9839
lag.quarterly.revenue 0.9978 1.0000 -0.9894 0.9817
price.index -0.9895 -0.9894 1.0000 -0.9539
income.level 0.9839 0.9817 -0.9539 1.0000
Variance inflation factor (<10 desired):
lag.quarterly.revenue price.index income.level
194.85 78.58 45.52
Standardized Residuals (observations > 3.00 problematic):
No significant outliers
Cook's distance (values >.2 problematic):
1963.25
0.8918
Normality of standardized model residuals: Shapiro-Wilk (p-value): 0.5586
Model change statistics
R R^2 Adj R^2 SE Est. Delta R^2 F Change df1 df2 p Fch Sig
Model 1 0.9978 0.9956 0.9955 0.0212 0.9956 8360.3793 1 37 0 ***
Model 2 0.9988 0.9977 0.9975 0.0159 0.0021 15.4599 2 35 0 ***
Model 1 : y ~ lag.quarterly.revenue
Model 2 : y ~ lag.quarterly.revenue + price.index + income.level
Model Coefficients
Model term estimate std.error statistic p.value sig
Model 1 (Intercept) 0.04169 0.10138 0.4112 0.6833
Model 1 lag.quarterly.revenue 0.99827 0.01092 91.4351 0.0000 ***
Model 2 (Intercept) 4.97077 1.24046 4.0072 0.0003 ***
Model 2 lag.quarterly.revenue 0.37305 0.11418 3.2673 0.0024 **
Model 2 price.index -0.81887 0.17152 -4.7742 0.0000 ***
Model 2 income.level 0.75435 0.14454 5.2189 0.0000 ***
