Trying to create a dataframe as result of a loop

Question

I'm trying to create a data frame that is a result of a loop:

variation <- seq(0.10, 3, 0.5)
for (i in seq_along(variation)) {
  x <- iris %>% mutate(newLength = Sepal.Length + variation[i])
  newSum <- x %>% summarise(newSum = sum(newLength))
  oldSum <- iris %>% summarise(oldSum = sum(Sepal.Length))

  df <- cbind(variation[i], oldSum, newSum)
  z <- rbind(df)
  print(z)
}

The output I'm getting is:

  variation[i] oldSum newSum
1          0.1  876.5  891.5
  variation[i] oldSum newSum
1          0.6  876.5  966.5
  variation[i] oldSum newSum
1          1.1  876.5 1041.5
  variation[i] oldSum newSum
1          1.6  876.5 1116.5
  variation[i] oldSum newSum
1          2.1  876.5 1191.5
  variation[i] oldSum newSum
1          2.6  876.5 1266.5

My desired output is:

 variation[i] oldSum newSum
         0.1  876.5  891.5
         0.6  876.5  966.5
         1.1  876.5 1041.5
         1.6  876.5 1116.5
         2.1  876.5 1191.5
         2.6  876.5 1266.5

What am I doing wrong?

Answer 1

rbind() binds multiple rows together. If you just give it one df, it will just return that data frame. Try rbind(z,df) to append the new DF to the old z.

variation <- seq(0.10, 3, 0.5)
for (i in seq_along(variation)) {
  x <- iris %>% 
    mutate(newLength = Sepal.Length + variation[i])
newSum <- x %>% 
    summarise(newSum = sum(newLength))
oldSum <- iris %>% 
  summarise(oldSum = sum(Sepal.Length))    
df <- cbind(variation[i], oldSum, newSum)
z <- rbind(z,df)
print(z)
}

Note that the z won't be cleared out, so you may wish to initialize it before you start your loop. Something like z = NULL would work to make sure it was empty.

Answer 2

You should try a vectorized function like outer to do the main complicated part of your analysis:

data.frame(
  variation,
  oldSum=sum(iris$Sepal.Length),
  newSum=colSums(outer(iris$Sepal.Length, variation, FUN=`+`))
)

#  variation oldSum newSum
#1       0.1  876.5  891.5
#2       0.6  876.5  966.5
#3       1.1  876.5 1041.5
#4       1.6  876.5 1116.5
#5       2.1  876.5 1191.5
#6       2.6  876.5 1266.5

As @Frank notes, you could simplify/speed this up even further:

sum.sl <- sum(iris$Sepal.Length)
data.frame(
  variation,
  oldSum=sum.sl,
  newSum=sum.sl + length(iris$Sepal.Length)*variation
)