问题
I have the following dataset:
>k1[1:10,]
id web_name first_name second_name position date team1 team2 game_week points home_away team_scored team_conceded minutes goals assists
1 1 Fabianski Lukasz Fabianski Goalkeeper 17/08/13 ARS AVL 1 0 H 1 3 0 0 0
2 1 Fabianski Lukasz Fabianski Goalkeeper 24/08/13 ARS FUL 2 0 A 3 1 0 0 0
3 1 Fabianski Lukasz Fabianski Goalkeeper 01/09/13 ARS TOT 3 0 H 1 0 0 0 0
4 1 Fabianski Lukasz Fabianski Goalkeeper 14/09/13 ARS SUN 4 0 A 3 1 0 0 0
5 1 Fabianski Lukasz Fabianski Goalkeeper 22/09/13 ARS STK 5 0 H 3 1 0 0 0
6 1 Fabianski Lukasz Fabianski Goalkeeper 28/09/13 ARS SWA 6 0 A 2 1 0 0 0
7 1 Fabianski Lukasz Fabianski Goalkeeper 06/10/13 ARS WBA 7 0 A 1 1 0 0 0
8 1 Fabianski Lukasz Fabianski Goalkeeper 19/10/13 ARS NOR 8 0 H 4 1 0 0 0
9 1 Fabianski Lukasz Fabianski Goalkeeper 26/10/13 ARS CRY 9 0 A 2 0 0 0 0
10 1 Fabianski Lukasz Fabianski Goalkeeper 02/11/13 ARS LIV 10 0 H 2 0 0 0 0
When I run the code:
> k2<- as.data.frame(k1 %>% group_by(id) %>%
mutate(min1 = lag(minutes, default=NA)) %>%
mutate(min3 = rollapply(min1,width=3,mean, align= "right", fill=NA)))
I receive the error:
Error: wrong sign in 'by' argument
However grouping by team1 via this code just works fine:
> k2<- as.data.frame(k1 %>% group_by(team1) %>%
mutate(min1 = lag(minutes, default=NA)) %>%
mutate(min3 = rollapply(min1,width=3,mean, align= "right", fill=NA)))
Pretty clueless why this is happening. Any help is appreciated.
回答1:
It can have different reasons, But simply it can happen if you pass a NULL dataframe to rollapply function
Also consider the using of zoo() as
rollapply(Zoo(min1),width=3,mean, align= "right", fill=NA)
来源：https://stackoverflow.com/questions/31896113/clueless-about-this-error-wrong-sign-in-by-argument