Add fitted quadratic curve

Question

I'm trying to add a fitted quadratic curve to a plot.

abline(lm(data~factor+I(factor^2)))

The regression which is displayed is linear and not quadratic and I get this message:

Message d'avis : In abline(lm(data ~ factor + I(factor^2)), col = palette[iteration]) : utilisation des deux premiers des 3 coefficients de régression

which means:

Use of the first 2 of the 3 regression coefficients

When running only the lm() function I don't get any messages.

Here is a sample data:

factor <- 1:7
data <- c(0.1375000,0.2500000,0.3416667,0.4583333,0.7250000,0.9166667,1.0000000)

score 11 · Accepted Answer · edited Jan 15 '14 at 01:49

11

Instead of using abline, use fitted, which gives you a vector the same length as your input of the predictions:

fitted(lm(data~factor+I(factor^2)))
#         1         2         3         4         5         6         7 
# 0.1248016 0.2395833 0.3699405 0.5158730 0.6773810 0.8544643 1.0471230

Thus, something like:

plot(factor, fitted(lm(data~factor+I(factor^2))), type="l")

edited Jan 15 '14 at 01:49

Uli Köhler

13,012
16
70
120

answered Feb 17 '13 at 22:52

David Robinson

77,383
16
167
187

I think it's helpful to know that this works more consistently (i.e., with numeric variables) when you `sort` both X and Y variables in `plot` (i.e., the original X value and the fitted data) . --> `plot(sort(x), sort(fitted(lm(y~x+I(x^2)))), type = 'l')` – theforestecologist Jan 25 '23 at 04:24

score 3 · Answer 2 · edited May 13 '19 at 10:58

I couldn't get answers so far to work, as dataset I used has x-values which are not increasing (as stated by David Robinson above). Here's how I solved it...

require(ISLR)
plot(mpg~horsepower, data=Auto)

# fit the model
glm.fit = glm(mpg~poly(horsepower,2), data=Auto)

# create 100 x-values based on min/max of plotted values
minMax = range(Auto$horsepower)
xVals = seq(minMax[1], minMax[2], len = 100) 

# Use predict based on a dataframe containing 'horsepower'
yVals = predict(glm.fit, newdata = data.frame(horsepower = xVals))

lines(xVals, yVals)

score 0 · Answer 3 · edited May 13 '19 at 10:59

0

thanks for all these valuable answer. Be carefull:

Use

Use predict based on a dataframe containing 'horsepower'

yVals = predict(glm.fit, newdata = data.frame(horsepower=xVals)

Use predict based on a dataframe containing 'horsepower'

yVals = predict(lm.fit, newdata = data.frame(horsepower=xVals)

lm.fit is a function

edited May 13 '19 at 10:59

MLavoie

9,671
41
36
56

answered Mar 12 '15 at 10:09

Souleymane

11
1

Add fitted quadratic curve

3 Answers3

Use predict based on a dataframe containing 'horsepower'

Use predict based on a dataframe containing 'horsepower'