CODEWITHSUNDEEP
rdplyr
jpinelo
jpinelo

Reputation: 1454

Pass function to data after group_by()

The following function extracts a linear regression model from a dataset. 

eqlabels <- function(data, x, y){
  m <- lm(x ~ y, data);
  eq <- substitute(italic(y) == a + b * italic(x) * "," ~~ italic(r) ^ 2 ~ "=" ~ r2, 
                   list(a = format(coef(m)[1], digits = 3),
                        b = format(coef(m)[2], digits = 3),
                        r2 = format(summary(m)$r.squared, digits = 2)))
  as.character(as.expression(eq));
  return(eq)
}

We can see it working 

s <- eqlabels(data = iris, x = iris$Sepal.Length, y = iris$Sepal.Width)
s

italic(y) == "6.53" + "-0.223" * italic(x) * "," ~ ~italic(r)^2 ~ 
    "=" ~ "0.014"

Question is how do I use this function with dplyr to calculate the equation and r2 values not for one group, but for several groups. For example:

result <- iris %>% group_by(Species) %>% eqlabels(x = iris$Sepal.Length, y = iris$Sepal.Width)

This seems to work but it only produces one, rather than three sets of equations and r2.

  > result
    italic(y) == "6.53" + "-0.223" * italic(x) * "," ~ ~italic(r)^2 ~ 
        "=" ~ "0.014"

?do seems to be a dplyr function for this but I can't get it to work...

result <- iris %>% group_by(Species) %>% do(eqlabels(x = iris$Sepal.Length, y = iris$Sepal.Width),.)

This expression halts the function...

Please note that I'm trying to avoid using ddply from the plyr package. Thank you

Upvotes: 1

Views: 75

Answers (1)

jeremycg
jeremycg

Reputation: 24945

Try:

result <- iris %>% group_by(Species) %>%
                   summarise(labels = list(eqlabels(., x = .$Sepal.Length, y = .$Sepal.Width)))

Source: local data frame [3 x 2]

     Species    labels
      (fctr)     (chr)
1     setosa <call[3]>
2 versicolor <call[3]>
3  virginica <call[3]>

If you want dplyr to work nicely with group_by, you need to pass it a mutate, summarise or do, rather than your own function. The output from your function is also not super nice - I've wrapped it in a list:

result$labels[[1]]
italic(y) == "6.53" + "-0.223" * italic(x) * "," ~ ~italic(r)^2 ~ 
    "=" ~ "0.014"

As the above comment mentions, you should use the broom package, it will make your life much easier.

Upvotes: 1

Related Questions