splinex.qmd

---
title: "Spline Examples"
date: last-modified
format:
  html:
    self-contained: true
execute:
  warning: false
  message: false

major: regression modeling
minor: ordinary least squares
---

```{r}
require(rms)
require(ggplot2)
options(prType='html')
n <- 100
set.seed(1)
x <- runif(100)
y <- abs(x - 0.5) + runif(100, -.2, .2)
real <- annotate('line', x=c(0, .5, 1), y=c(.5, 0, .5), col='red', alpha=0.4)
# Need mapping= since no data before aes
ggplot(mapping=aes(x, y)) + geom_point() + real
# Could also do plot(x, y)
dd <- datadist(x); options(datadist='dd')
# Fit the true model
f <- ols(y ~ abs(x - 0.5))
ggplot(Predict(f)) + real
# Try various models: linear, polynomial, restricted cubic spline, linear spline
f <- ols(y ~ x)
ggplot(Predict(f)) + real
specs(f)
latex(f)
```