如何随机选择2个实例（2行）进行测试＆amp;留在数据集中进行培训

Question

如何随机选择2个实例（2行）进行测试＆amp;留在像这样的样本数据集中进行培训。

dog.data

22.0,4566.0,56.4,89.3,Dog-fifota
81.0,3434.0,34.4,67.3,Dog-listem
30.0,8944.0,23.4,45.3,Dog-biker
55.0,3455.0,78.5,11.3,Dog-listem
41.4,3345.0,45.3,34.1,Dog-fifota

Answer 1

尝试（在R中）

indx <- sample(nrow(dog.data), 2) 
test <- dog.data[indx, ] 
train <- dog.data[-indx, ]

---

修改

如果你想把它作为一个函数，这样的东西就可以了：

spltfunc <- function(x){
  indx <- sample(nrow(x), 2)
  test <- x[indx, ] 
  train <- x[-indx, ]
  list2env(list(test = test, train = train), .GlobalEnv)
}

测试

set.seed(123) # setting the random seed so you can reproduce results
spltfunc(dog.data)
# <environment: R_GlobalEnv>
test
#   V1   V2   V3   V4         V5
# 2 81 3434 34.4 67.3 Dog-listem
# 4 55 3455 78.5 11.3 Dog-listem
train
#     V1   V2   V3   V4         V5
# 1 22.0 4566 56.4 89.3 Dog-fifota
# 3 30.0 8944 23.4 45.3  Dog-biker
# 5 41.4 3345 45.3 34.1 Dog-fifota

Answer 2

在随机样本和其他样本之间拆分数据集的一种简单方法是对整个数据集进行混洗，然后进行切片。您将获得两个随机排序的项目列表。

这是一个Python函数：

import random

def split_test_training(data, num_test_rows):
    data_copy = list(data)           # shallow copy
    random.shuffle(data)             # reorder the copied list in-place

    testing = data[:num_test_rows]   # slice the first num_test_rows for testing
    training = data[num_test_rows:]  # slice the rest for training

    return testing, training

如果您不介意传入的列表被该功能改组，您可以跳过浅层副本。