CODEWITHSUNDEEP
rdplyr
Diego
Diego

Reputation: 2356

Apply my own function to dplyr tbl_df

Time ago I've made this functions:

vap.vector <- function (x, y) {
    y[is.infinite(y)] <- NA
    y[y == 0] <- NA
    logtasa <- log(y)
    datos <- data.frame(x, logtasa)
    datos <- na.omit(datos)
    lmdatos <- lm(logtasa ~ x, data = datos)
    es <- qt(0.975, lmdatos$df.residual) * summary(lmdatos)$coefficients[2, 2]
    vap <- summary(lmdatos)$coefficients[2, 1]
    vec <- c(vap = vap, es = es, lwr = vap - es, upr = vap + es)
    round((1 - exp(vec)) * -100, 2)
}

vap_ <- function (df, x, y) {
    x = lazyeval::lazy_eval(x, df)
    y = lazyeval::lazy_eval(y, df)
    vap.vector(x, y)
}

vap <- function (df, x, y) 
vap_(df, lazyeval::lazy(x), lazyeval::lazy(y))

I have this dataset

https://drive.google.com/file/d/0Bw2XRcfksYZuX1R5bG9oZzdhZ2M/view?usp=sharing

and I wanna do:

tbl_df %>% group_by(code, sex) %>% vap(year, values)

or 

tbl_df %>% group_by(code, sex) %>%
           summarise(vap = vap.vector(year, values)[1],
                     lwr = vap.vector(year, values)[3],
                     upr = vap.vector(year, values)[4]
 )

and as result a data.frame with code, sex, and vap, lwr and upr columns.

but I cannot do it. I the second aproach I get:

Error: 0 (non-NA) cases

There is a simply way to do it or I need to be @hadley

I'm working on Windows with R 3.2.2 and dplyr 0.4.3. Thanks in advance...

Upvotes: 1

Views: 268

Answers (1)

tguzella
tguzella

Reputation: 1481

You just need do use do:

vap.vector <- function (x, y) {
    y[is.infinite(y)] <- NA
    y[y == 0] <- NA
    logtasa <- log(y)
    datos <- data.frame(x, logtasa)
    datos <- na.omit(datos)
    lmdatos <- lm(logtasa ~ x, data = datos)
    es <- qt(0.975, lmdatos$df.residual) * summary(lmdatos)$coefficients[2, 2]
    vap <- summary(lmdatos)$coefficients[2, 1]
    vec <- c(vap = vap, es = es, lwr = vap - es, upr = vap + es)
    round((1 - exp(vec)) * -100, 2)
}

Data <- read.table("dataset.csv", sep = ",", header = TRUE)

library(dplyr)
Results <- 
    Data %>% 
    group_by(code, sex) %>% 
    do(as.data.frame(t(vap.vector(.$year, .$values)[-2])))

which produces:

> head(as.data.frame(Results))
  code sex  vap  lwr  upr
1    0   F 2.10 1.07 3.13
2    0   M 0.92 0.00 1.85
3    2   F 3.82 2.91 4.73
4    2   M 2.72 1.95 3.49
5    4   F 3.23 2.25 4.21
6    4   M 3.76 2.70 4.83

Upvotes: 1

Related Questions