r_help.tex

\begin{codebox}{r}{R-Help}
# Workaround for automatic dimension collapse
matrix(X[test.fold,], nrow=n/k)

# Test if an element is in a list
if ("X1" %in% names(coef(m,mo)))

# Fit distribution to data vector
library(MASS)
fit.gamma <- fitdistr(boogg, "gamma")

# Remove NA
thuesen <- thuesen[!is.na(thuesen[,2]),]
data_comp <- data[complete.cases(data),]
# Creating categorical variables
High=ifelse(Carseats$Sales<=8,"No","Yes")
# Standardize data
scaled.dat <- scale(dat)

# Anova test to determine if there is a significant
# difference between models. Anova uses RSS and DoF
# of largest (last) model, so use ascending order!
anova(fit.0, fit.1, fit.2, fit.3)

# Loess smoother
span <- c(0.1, 0.2, 0.3, 0.45, 0.7, 1)
lo <- loess(y ~ x, span=span[i])
prediction <- predict(object=lo, x)

# Given fixed x, error distribution and true param.:
# Power of test simulation. Know that y ~ poly(x, 3) + err (for typeI error: do same with y = err)
results.power <- numeric(n.sim)
for (i in 1:n.sim) {
    err <- rgamma(n, ...) - 2
    y <- beta.0 + beta.1 * I(x) + beta.2 * I(x^2) + beta.3 + I(x^3) + err
    fit.power <- lm(y ~ I(x) + I(x^2) + I(x^3))
    f1 <- summary(fit.power)$fstatistic
    p.val.power <- 1 - pf(f1[1], f1[2], f1[3])
    results.power[i] <- p.val.power < 0.05
}
power <- mean(results.power)

# Plotting
par(mfrow=c(1,1)) # Arrangement of plots
plot(..., xlim=c(-2,2), ylim=c(-5,8), xlab="x", ylab="y")
abline(a=..., b=..., col="red") # Add straight line
abline(h=..., col="red") # Add horizontal line
abline(v=..., col="red") # Add vertical line
lines(density(some.vector)) # Add density
# draw density and CDF
grid <- seq(from=0,to=5,length=200)
plot(grid, dlnorm(grid), type="l", main="density")
plot(grid, plnorm(grid), type="l", main="CDF")
# Add regression line to data plot
library(MASS)
attach(Boston)
fit <- lm(dis ~ poly(nox, degree = 3))
line.x <- seq(min(nox), max(nox), length.out = 1000)
line.y <- predict(fit, data.frame(nox = line.x))
plot(dis, nox)
lines(line.y, line.x, col = "red")
\end{codebox}