清洁数据和关联规则-R [英] Cleaning Data & Association Rules - R

问题描述

我正在尝试整理R中的以下数据集（在链接中），然后在下面运行关联规则。

I am trying to tidy the following dataset (in link) in R and then run an association rules below.

https://www.kaggle.com/fanatiks/shopping-cart

install.packages("dplyr")
library(dplyr)

df <- read.csv("Groceries (2).csv", header = F, stringsAsFactors = F, na.strings=c(""," ","NA"))
install.packages("stringr")
library(stringr)
temp1<- (str_extract(df$V1, "[a-z]+"))
temp2<- (str_extract(df$V1, "[^a-z]+"))
df<- cbind(temp1,df)
df[2] <- NULL
df[35] <- NULL
View(df)

summary(df)
str(df)

trans <- as(df,"transactions")

当我运行上面的trans<-as（df，交易）代码时，出现以下错误：

I get the following error when I run the above trans <- as(df,"transactions") code:

警告消息：
列2、3、4、5、6、7、8、9、10、11、12、13、14、15、16、17、18、19、20、21、22， 23、24、25、26、27、28、29、30、31、32、33、34不是逻辑或因数。应用默认离散化（请参见'？DiscretizeDF'）。

Warning message: Column(s) 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34 not logical or factor. Applying default discretization (see '? discretizeDF').

summary(trans)

运行上面的代码，我得到以下信息：

When I run the above code, I get the following:

transactions as itemMatrix in sparse format with
 1499 rows (elements/itemsets/transactions) and
 1268 columns (items) and a density of 0.01529042 

most frequent items:
  V5= vegetables   V6= vegetables temp1=vegetables   V2= vegetables 
             140              113              109              108 
  V9= vegetables          (Other) 
             103            28490 

所附结果显示所有蔬菜价值作为单独的项目显示，而不是组合蔬菜评分，这显然增加了我的专栏数。我不确定为什么会这样吗？

The attached results is showing all the vegetable values as separate items instead of a combined vegetable score which is obviously increasing my number of columns. I am not sure why this is happening?

fit<-apriori(trans,parameter=list(support=0.006,confidence=0.25,minlen=2))
fit<-sort(fit,by="support")
inspect(head(fit))

推荐答案

要强制转换为事务类，数据框必须由因子列组成。您有一个字符数据框-因此出现错误消息。数据需要进一步清理才能使其正确强制。

For coercion to transaction class the dataframe needs to be made up of factor columns. You have a dataframe of characters - hence the error message. The data requires some further cleaning in order to get it to coerce properly.

我对arules软件包不是很熟悉，但我相信read.transactions函数可能是更有用，因为它会自动丢弃重复项。我发现制作二进制矩阵并使用for循环是最容易的，但是我确信有一个更整洁的解决方案。

I'm not very familiar with the arules package but I believe the read.transactions function may be more useful as it would automatically discard duplicates. I found it easiest to make a binary matrix and use a for loop, but I am sure there is a neater solution.

直接从您的代码继续：

items <- as.character(unique(unlist(df))) # get all unique items
items <- items[which(str_detect(items, "[a-z]"))] # remove numbers


trans <- matrix(0, nrow = nrow(df), ncol = length(items))


for(i in 1:nrow(df)){
  trans[i,which(items %in% t(df[i,]))] <- 1
}

colnames(trans) <- items
rownames(trans) <- temp2

trans <- as(trans, "transactions")

summary(trans)

给予

transactions as itemMatrix in sparse format with
 1637 rows (elements/itemsets/transactions) and
 38 columns (items) and a density of 0.3359965 

most frequent items:
 vegetables     poultry     waffles   ice cream  lunch meat     (Other) 
       1058         582         562         556         555       17588 

element (itemset/transaction) length distribution:
sizes
  0   1   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26 
102  36   8  57  51  51  71  69  63  80  79  58  84  91  72 105  97  87 114  91  82  46  30   7   4   2 

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
   0.00    8.00   14.00   12.77   18.00   26.00 

includes extended item information - examples:
    labels
1     pork
2  shampoo
3    juice

includes extended transaction information - examples:
  transactionID
1      1/1/2000
2      1/1/2000
3      2/1/2000

这篇关于清洁数据和关联规则-R的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！