我是R的新手并且学习基本的东西。我在R中有一个数据框，其中包含controller_id，user_id，mth_id，col_val1等列，直到col_val100。
df <- data.frame('controller_id' = c('X','X','X','X','X','X','Y','Y','Y','Y','Y','Y','Z','Z'),
'user_id'=c('A','B','C','A','B','C','P','Q','R','P','Q','R',NA,NA),
'mth_id'=c('1393','1393','1393','1398','1398','1398','1393','1393','1393','1398','1398','1398','1393','1398'),
'col_val1' = c(5,4,6,3,1,10,12,15,18,13,19,1,5,2),
'col_val2'=c(8,12,9,2,12,5,7,9,11,4,0,7,10,5))
> df
controller_id user_id mth_id col_val1 col_val2
1 X A 1393 5 8
2 X B 1393 4 12
3 X C 1393 6 9
4 X A 1398 3 2
5 X B 1398 1 12
6 X C 1398 10 5
7 Y P 1393 12 7
8 Y Q 1393 15 9
9 Y R 1393 18 11
10 Y P 1398 13 4
11 Y Q 1398 19 0
12 Y R 1398 1 7
13 Z <NA> 1393 5 10
14 Z <NA> 1398 2 5
我想要的是根据特定的user_id，mth_id计算每个controller_id的col_values的差异，并仅显示随着mth_id的增加而减少的col_values。
例如：对于controller_id = X，我们有3个user_id作为A，B，C用于两个不同的mth_id。代码应该为所有3个users_id计算m__id 1398和1393的col_val1之间的差异，并且如果该差异是＆lt; 0然后我想要像
这样的输出Col_val1 for controller_id 'X', user_id 'A' has decreased from 5 to 3
如果给定的controller_id没有关联的user_id，那么它应该计算controller_id本身之间的列值差异。
理想情况下，我希望将这些输出存储在列表/数据框中以供以后使用。 此外，代码将运行约。数据框中有900列。
非常感谢任何帮助。
答案 0 :(得分：1)
考虑使用运行组总和的基本R解决方案。要迭代所有列，请使用sapply()传递列名：
rowdiff <- function(col) {
sapply(1:nrow(df),
function(i){
# CONDITIONAL TO RETURN NA FOR FIRST VAL IN EACH USER ID
ifelse(sum(df[1:i, c("user_id")] == df$user_id[i]) == 1, NA,
# DIFFERENCE OF CURRENT LOOP COL VALUE - LAST COL VALUE OF USER ID GROUP
df[[col]][i] -
sum((df[1:i-1, c("user_id")] == df$user_id[i])
* df[1:i-1,][[col]]))
})
}
finaldf <- cbind(df, data.frame(sapply(names(df[c(3:ncol(df))]), rowdiff)))
# user_id mth_id col_val1 col_val2 col_val3 col_val1 col_val2 col_val3
# 1 A 1398 4 2 12 NA NA NA
# 2 B 1398 3 3 30 NA NA NA
# 3 C 1398 1 1 14 NA NA NA
# 4 A 1393 5 7 7 1 5 -5
# 5 B 1393 2 6 18 -1 3 -12
# 6 C 1393 7 0 9 6 -1 -5
# 7 D 1398 4 5 12 NA NA NA
# 8 D 1393 0 3 24 -4 -2 12
你是否需要写出来的陈述：
statements <- function(col) {
sapply(1:nrow(df),
function(i){
delta <- df[[col]][i]-
sum((df[1:i-1, c("controller_id")] == df$controller_id[i])
*(df[1:i-1, c("user_id")] == df$user_id[i])
* df[1:i-1,][[col]])
changeword <- ifelse(delta < 0, "decreased",
ifelse(delta > 0, "increased", "not changed"))
ifelse(sum(df[1:i, c("user_id")] == df$user_id[i]) == 1, NA,
paste0(col, " for controller_id '", df$controller_id[i], "', user_id '",
df$user_id[i], "' has ", changeword, " from ",
sum((df[1:i-1, c("controller_id")] == df$controller_id[i])
* (df[1:i-1, c("user_id")] == df$user_id[i])
* df[1:i-1,][[col]]), " to ",
df[[col]][i])
)
})
}
finaldf <- cbind(df, data.frame(sapply(names(df[c(4:ncol(df))]), statements)))
输出
col_val1
1 <NA>
2 <NA>
3 <NA>
4 col_val1 for controller_id 'X', user_id 'A' has decreased from 5 to 3
5 col_val1 for controller_id 'X', user_id 'B' has decreased from 4 to 1
6 col_val1 for controller_id 'X', user_id 'C' has increased from 6 to 10
7 <NA>
8 <NA>
9 <NA>
10 col_val1 for controller_id 'Y', user_id 'P' has increased from 12 to 13
11 col_val1 for controller_id 'Y', user_id 'Q' has increased from 15 to 19
12 col_val1 for controller_id 'Y', user_id 'R' has decreased from 18 to 1
13 <NA>
14 <NA>
col_val2
1 <NA>
2 <NA>
3 <NA>
4 col_val2 for controller_id 'X', user_id 'A' has decreased from 8 to 2
5 col_val2 for controller_id 'X', user_id 'B' has not changed from 12 to 12
6 col_val2 for controller_id 'X', user_id 'C' has decreased from 9 to 5
7 <NA>
8 <NA>
9 <NA>
10 col_val2 for controller_id 'Y', user_id 'P' has decreased from 7 to 4
11 col_val2 for controller_id 'Y', user_id 'Q' has decreased from 9 to 0
12 col_val2 for controller_id 'Y', user_id 'R' has decreased from 11 to 7
13 <NA>
14 <NA>