CODEWITHSUNDEEP
rnasmoothing
Kacey
Kacey

Reputation: 147

R: Function "lowess" with NAs

I am trying to use function "lowess" in base package to smooth a vector. The question is that the vector only has one NA value, but after smoothing by "lowess", 4 more NAs appear. I searched and some one suggest using "lowess" in package "gplots". I tried but got same results.

x1 <- c(NA, 19.93621, 17.64789, 17.78488, 17.11141, 18.4648, 19.62629, 17.5737, 17.48582, 18.76946, 19.57138, 19.62812, 2.982385, -0.1320513)
x2 <- lowess(x1)
x2
$x
[1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14

$y
[1]        NA        NA        NA        NA        NA 18.988279 18.563642 18.407768 18.496699 17.922510 14.419999 10.861535  7.179754  3.145612

One way is to delete the NA in x1 so there is no NA after smoothing in x2$y.

x2 <- lowess(x1[-1])
x2
$x
 [1]  1  2  3  4  5  6  7  8  9 10 11 12 13

$y
 [1] 18.7079309 18.4481273 18.2491632 18.0847709 18.0946245 18.1282192 18.1497617 17.9298278 14.6441882 10.9465210  7.2635244  3.5247529 -0.3021372

But I just wonder why more NAs appear and is there an easier way? Thank you!

Upvotes: 3

Views: 2604

Answers (1)

Glen_b
Glen_b

Reputation: 8252

lowess doesn't have any built-in NA handling

  1. Probably the easiest thing to do long term is learn to use loess (whose options and settings are slightly different), which does deal with NA values (it has an na.action argument, like lm does, which by default should work much the way you want - however, in terms of syntax, loess works much more like lm than it does like lowess).

  2. you can use na.omit directly on the arguments to lowess; that way you don't need to specifically identify which observations are omitted.

Upvotes: 4

Related Questions