confusionMatrix.matrix(data = df_name, reference = numeric_vector_name, positive = "increase") 错误：矩阵必须具有相等的维度

Question

我已经连续 3 次在这篇文章的标题中出现错误，当我将此错误消息输入 Google 或 Bing 时，在 Stack Overflow 上出现的前几个类似问题都没有适合我的特定情况的答案。 以下代码行来自的脚本位于我的GitHub 存储库中，该分析项目位于名为“AIT 622 大数据分析项目脚本”的 R 脚本文件中。

### Classification Forecasting Model #5: Multivariate Adaptive Regression Splines
library(earth)
library(plotmo)
library(plotrix)
marsGrid = expand.grid(.degree = 1:2, .nprune = 2:38)
set.seed(100)
marsModelR1 = train(x = data2014, y = pr_var2014, method = "earth", 
                    preProc = c("center", "scale"), tuneGrid = marsGrid)

# compare the expected classifications in 2015 to the observed classifications in 2015
marsR1Pred = predict(marsModelR1, newdata = data2015)
> dim(marsR1Pred)
[1] 4120    1

我在此处添加以下内容以显示我认为可能是错误消息/警告的来源。 > 长度（pr_var2014） 1 3808

marsR1_PR = postResample(pred = marsR1Pred, obs = pr_var2014)
> marsR1_PR
       RMSE    Rsquared         MAE 
         NA 1.24489e-06          NA

marsModelR1_CFM <- confusionMatrix(data = marsR1Pred, reference = pr_var2014,
                                   positive = "Increase")
> marsModelR1_CFM <- confusionMatrix(data = marsR1Pred, reference = pr_var2014,
+                                    positive = "Increase")
Error in confusionMatrix.matrix(data = marsR1Pred, reference = pr_var2014,  : 
  matrix must have equal dimensions

我已经尝试了以下 2 次尝试修复，两者的结果相同

> marsModelR1_CFM <- confusionMatrix(data = marsR1Pred, 
+                                    reference = sample(pr_var2014, length(marsR1Pred)),
+                                    positive = "Increase")
Error in confusionMatrix.matrix(data = marsR1Pred, reference = sample(pr_var2014,  : 
  matrix must have equal dimensions

> length(pr_var2014)
[1] 3808
> length(marsR1Pred)
[1] 4120
> marsModelR1_CFM <- confusionMatrix(data = marsR1Pred, 
+                                    reference = sample(pr_var2014, 4120),
+                                    positive = "Increase")
Error in confusionMatrix.matrix(data = marsR1Pred, reference = sample(pr_var2014,  : 
  matrix must have equal dimensions

任何建议将不胜感激，此外，所有这些命令都在我的脚本的最后部分。 第 4 部分在底部。

Answer 1

确保您使用的是数据参数的因子，而不是matrix 。

# this will fail with similar error to yours
confusionMatrix(as.matrix(iris$Species), sample(iris$Species))
# Error in confusionMatrix.matrix(as.matrix(iris$Species), sample(iris$Species)) : 
#  matrix must have equal dimensions

# this will pass as I convert the variable to be a factor variable
confusionMatrix(as.factor(as.matrix(iris$Species)), sample(iris$Species))

对于你的情况：

marsModelR1_CFM <- confusionMatrix(data = as.factor(marsR1Pred), 
                                   reference = sample(as.factor(pr_var2014), 4120),
                                   positive = "Increase")