“火车”和“类” R 中的长度不同

Question

我正在做一些测试工作。我面临以下错误：

Error in knn(train = Data_train, test = Data_test, cl = Data_train_label,  : 
  'train' and 'class' have different lengths

可重现的代码：

# Load data
library(data.table)

Data = fread('https://raw.githubusercontent.com/justmarkham/DAT5/master/data/auto_mpg.txt', stringsAsFactors = FALSE)
Data

str(Data)

# K-mean

Data$cylinders = factor(Data$cylinders, levels = c('3', '4', '5', '6', '8'), label = c('3_cylinders', '4_cylinders', '5_cylinders', 
                                                                                       '6_cylinders', '8_cylinders'))

round(prop.table(table(Data$cylinders))*100, digits = 1)

Data = Data[, c('cylinders', 'mpg', 'horsepower', 'displacement', 'weight', 'acceleration', 'model_year', 'origin', 'car_name')]
head(Data)

summary(Data[, c('mpg', 'horsepower')])

normalize = function(x) {return ((x-min(x) / max(x) - min(x)))}


Data_n = as.data.frame(lapply(Data[, 2:8], normalize))

smp_size = floor(0.8 * nrow(Data))
set.seed(123)

train_ind = sample(seq_len(nrow(Data)), size = smp_size)

Data_train = Data_n[train_ind, ]
Data_test = Data_n[-train_ind, ]

dim(Data_train)
dim(Data_test)

Data_train_label = Data[train_ind, 1]
Data_test_label = Data[-train_ind, 1]

length(Data_train_label)
length(Data_test_label)


#install.packages("class")

library('class')

Data_test_pred = knn(train = Data_train, test = Data_test, cl = Data_train_label, k=19)

#install.packages("gmodels")

library(gmodels)
CrossTable(x = Data_test_label$cylinders, y= Data_test_pred, prop.chisq=FALSE)

显然，问题是在检查长度时，我得到以下信息：

> length(Data_train_label)
[1] 1
> length(Data_test_label)
[1] 1

并且应该收到：

> length(Data_train_label)
[1] 313
> length(Data_test_label)
[1] 79

这很奇怪，我查看了有关此主题的其他问题，但没有找到任何内容这可以帮助我。也许Data_train_label需要转换成向量？

Answer 1

您的类的参数 cl 需要一个向量而不是数据帧。

以下代码应该可以工作：

Data_test_pred = knn(
  train = Data_train, test = Data_test,
  cl = as.vector(as.matrix(Data_train_label)), # Convert labels to vector
  k=19
)

Answer 2

这对我来说是工作

Data_train_label = Data[train_ind, 1]
Data_test_label = Data[-train_ind, 1]

length(Data_train_label$cylinders)
length(Data_test_label$cylinders)


#install.packages("class")

library('class')

Data_test_pred = knn(train = Data_train, test = Data_test, cl = Data_train_label$cylinders, k=19)