来自spmf的关联规则中的R子集

时间:2014-08-01 08:36:29

标签: r subset apriori

代码:

data=read.csv("./spmf1234.csv",header=F);

df<- data.frame(do.call('rbind', strsplit(as.character(data$V1),'#',fixed=F)));

df2<- data.frame(do.call('rbind', strsplit(as.character(df$X1),'==>',fixed=F)));

df3=cbind(df2,df);

colnames(df3)=c("lhs","rhs","rule","support","confidence","lift");

df4 <- subset(df3, select = c(lhs,rhs,support,confidence,lift));

final=subset(df4,lhs!=1);

我正在使用fp-growth从spmf获取的csv文件上尝试上面的代码来获取关联规则。我希望所有规则都包含&#39; 1&#39;在lhs被删除,但这不起作用。

csv文件:

2 ==&gt; 1 #SUP:1 #CONF:0.33333 #LIFT:0.66667

1 ==&gt; 2 #SUP:1 #CONF:0.33333 #LIFT:0.66667

3 ==&gt; 1 #SUP:2 #CONF:0.5 #LIFT:1

1 ==&gt; 3 #SUP:2 #CONF:0.66667 #LIFT:1

3 ==&gt; 2 #SUP:2 #CONF:0.5 #LIFT:1

2 ==&gt; 3 #SUP:2 #CONF:0.66667 #LIFT:1

2 3 ==&gt; 1 #SUP:1 #CONF:0.5 #LIFT:1

1 3 ==&gt; 2 #SUP:1 #CONF:0.5 #LIFT:1

1 2 ==&gt; 3 #SUP:1 #CONF:1 #LIFT:1.5

3 ==&gt; 1 2 #SUP:1 #CONF:0.25 #LIFT:1.5

2 ==&gt; 1 3 #SUP:1 #CONF:0.33333 #LIFT:1

1 ==&gt; 2 3 #SUP:1 #CONF:0.33333 #LIFT:1

1 个答案:

答案 0 :(得分:1)

final=subset(df4,lhs!=1) 

将字符与数字进行比较,但不起作用:

# > as.character(df4$lhs)
# [1] "2 "   "1 "   "3 "   "1 "   "3 "   "2 "   "2 3 " "1 3 " "1 2 " "3 "   "2 "   "1 "  

你可能想使用像这样的一个正则表达式:

final = subset(df4, !grepl("1\\b", lhs))
# > final
#    lhs   rhs support     confidence          lift
# 1    2     1  SUP: 1  CONF: 0.33333  LIFT: 0.66667
# 3    3     1  SUP: 2      CONF: 0.5        LIFT: 1
# 5    3     2  SUP: 2      CONF: 0.5        LIFT: 1
# 6    2     3  SUP: 2  CONF: 0.66667        LIFT: 1
# 7  2 3     1  SUP: 1      CONF: 0.5        LIFT: 1
# 10   3   1 2  SUP: 1     CONF: 0.25      LIFT: 1.5
# 11   2   1 3  SUP: 1  CONF: 0.33333        LIFT: 1

添加:

## data preperation
data <- readLines(con = textConnection("  
2 ==> 1 #SUP: 1 #CONF: 0.33333 #LIFT: 0.66667
1 ==> 2 #SUP: 1 #CONF: 0.33333 #LIFT: 0.66667
3 ==> 1 #SUP: 2 #CONF: 0.5 #LIFT: 1
1 ==> 3 #SUP: 2 #CONF: 0.66667 #LIFT: 1
3 ==> 2 #SUP: 2 #CONF: 0.5 #LIFT: 1
2 ==> 3 #SUP: 2 #CONF: 0.66667 #LIFT: 1
2 3 ==> 1 #SUP: 1 #CONF: 0.5 #LIFT: 1
1 3 ==> 2 #SUP: 1 #CONF: 0.5 #LIFT: 1
1 2 ==> 3 #SUP: 1 #CONF: 1 #LIFT: 1.5
3 ==> 1 2 #SUP: 1 #CONF: 0.25 #LIFT: 1.5
2 ==> 1 3 #SUP: 1 #CONF: 0.33333 #LIFT: 1
1 ==> 2 3 #SUP: 1 #CONF: 0.33333 #LIFT: 1"))
r <- regexec(pattern = "([0-9 ]+)\\s==>\\s([0-9 ]+)\\s\\#SUP:\\s([0-9.]+)\\s\\#CONF:\\s([0-9.]+)\\s\\#LIFT:\\s([0-9.]+)", 
             text = data)
m <- regmatches(data, r)
df <- setNames(as.data.frame(do.call(rbind, lapply(m, "[", -1)), stringsAsFactors = FALSE), 
               c("lhs", "rhs", "support", "confidence", "lift"))

## rows to include/exclude
include <- sapply(strsplit(df$lhs, " "), function(x) !any(as.integer(x) %in% 1:1000))
df[include, ]