【R语言】关联规则

#数据集
install.packages("arules")
library(arules)
data(Groceries)
summary(Groceries)
inspect(Groceries[1:5])

itemFrequency(Groceries[,1:3])
# frankfurter     sausage  liver loaf 
# 0.058973055 0.093950178 0.005083884 
itemFrequencyPlot(Groceries,support=0.1)
itemFrequencyPlot(Groceries,topN=20)

训练模型

#构建模型
freq.itemsets <- eclat(Groceries,parameter = list(supp=0.075,maxlen=15))
inspect(freq.itemsets)
#      items              support    count
# [1]  {whole milk}       0.25551601 2513 
# [2]  {other vegetables} 0.19349263 1903 
# [3]  {rolls/buns}       0.18393493 1809 
# [4]  {yogurt}           0.13950178 1372 
# [5]  {soda}             0.17437722 1715 
# [6]  {root vegetables}  0.10899847 1072 
# [7]  {tropical fruit}   0.10493137 1032 
# [8]  {bottled water}    0.11052364 1087 
# [9]  {sausage}          0.09395018  924 
# [10] {shopping bags}    0.09852567  969 
# [11] {citrus fruit}     0.08276563  814 
# [12] {pastry}           0.08896797  875 
# [13] {pip fruit}        0.07564820  744 
# [14] {newspapers}       0.07981698  785 
# [15] {bottled beer}     0.08052872  792 
# [16] {canned beer}      0.07768175  764 

#创建规则
groceryrules <- apriori(Groceries, parameter = list(support =0.006, confidence = 0.25, minlen = 2))
groceryrules                        
#set of 463 rules                        

模型的评估

#模型的评估
summary(groceryrules)
inspect(groceryrules[1:3])

提升关联规则的效果

#检查关联规则集的有效方法是查看具有更高提升度的规则，因为较大的提升度表示先前隐藏在项之间的强连接。
inspect(sort(groceryrules,by="lift")[1:5])

#检查数据规则的另一种有效方法是查看包含某特定规则子集。
#假如要发现包含berries的集合，实现方式是使用subset筛选出items包含berries的项目。
#如果要求一个人在特定季节制作促销berries的广告，这样的操作将非常有用。
berryrules <- subset(groceryrules,items %in% "berries")
inspect(berryrules)

关联规则的可视化

#关联规则的可视化
install.packages("arulesViz")
library(arulesViz)
library(grid)
plot(berryrules,method = "graph")
#圆圈颜色越深，表示提升度越高；圆圈越大，表示这条规则出现的次数越多。

#treemap图
install.packages("tidyverse")
install.packages("treemap")
library(tidyverse)
library(treemap)
occur1 <- Groceries@itemInfo %>% group_by(level1) %>%summarize(n=n())
occur2 <- Groceries@itemInfo %>% group_by(level1,level2) %>%summarize(n=n())
occur3 <- Groceries@itemInfo %>% group_by(level1,level2,labels) %>%summarize(n=n())
treemap(occur1,index = c("level1"),vSize = "n")
treemap(occur2,index = c("level1","level2"),vSize = "n")
treemap(occur3,index = c("level1","level2","labels"),vSize = "n")