Predicting new values from a model.

Question

[english.server]

Very basic question here (so that I'm afraid to ask). Suppose I have 50 pairs of data and I want to linearly model my data and predict a single new pair to my single input. What should I do?

a= matrix (rnorm(100),50)
l= lm (a[,2]~a[,1])
predict (l, 2.1)

doesn't work.

Answer 1

[zx8754]

Try this example:

# reproducible example input
set.seed(1); a <- data.frame(matrix(rnorm(100), ncol = 2))
head(a)
#           X1         X2
# 1 -0.6264538  0.3981059
# 2  0.1836433 -0.6120264
# 3 -0.8356286  0.3411197
# 4  1.5952808 -1.1293631
# 5  0.3295078  1.4330237
# 6 -0.8204684  1.9803999

# get fit
l <- lm (X2 ~ X1, data = a)

# new input data with only X1 column, because we want to predict X2
newData <- data.frame(X1 = 2.1)

newData$X2 <- predict(l, newData)
newData
#    X1         X2
# 1 2.1 0.02624987

[english.server]

Thank you indeed very much.