(from Wikipedia) Akaike information criterion (AIC) is a measure of the relative quality of a statistical model for a given set of data. That is, given a collection of models for the data, AIC estimates the quality of each model, relative to the other models. Hence, AIC provides a means for model selection. It is formally defined as:
where L be the maximized value of the likelihood function ( 
AIC gives us an opportunity to compare a set of models, and evaluate a set of parameters, relative to each other. We learn nothing about the quality of the models themselves, but can at least identify which among a set of models warrant further investigation.
The example below uses the Kaggle Titanic dataset. It creates 3 different linear models with different parameter sets to see how AIC changes. This is a very basic example.
> trainDF <- read.csv('train.csv')
> testLM <- lm(Survived ~ Age + SibSp + Pclass + Sex,
data=subset(trainDF, !is.na(Age) & !is.na(SibSp)))
> AIC(testLM)
[1] 875.0388
> testLM_1 <- lm(Survived ~ Age + Pclass,
data=subset(trainDF, !is.na(Age)))
> testLM_2 <- lm(Survived ~ Age + Pclass + Sex,
data=subset(trainDF, !is.na(Age)))
> AIC(testLM, testLM_1, testLM_2)
df AIC
testLM 6 658.4335
testLM_1 4 876.7700
testLM_2 5 667.707
The results of the AIC comparison suggest that the parameters in testLM_2 are the best set to move forward with.
There is also a stepwise AIC finctional capability contained within the MASS package that may be of some use, though I have not had an opportunity to explore its meaning
> library(MASS)
> step <- stepAIC(testLM, direction="both")
Start: AIC=-1369.81
Survived ~ Age + SibSp + Pclass + Sex
Df Sum of Sq RSS AIC
<none> 103.38 -1369.8
- SibSp 1 1.645 105.02 -1360.5
- Age 1 4.971 108.35 -1338.3
- Pclass 1 17.439 120.82 -1260.5
- Sex 1 37.031 140.41 -1153.2
[1] R Documentation regarding AIC
[3] Quick-R Multiple Linear Regression Page, section re: Variable Selection