The following code produces different outcomes when Run -> Run All is executed and when Knit is executed. Any comments are appreciated, thank you.
---
title: "Ch8-Bayessian Additive Regression Trees-lab"
jupyter:
jupytext:
formats: ipynb,Rmd
text_representation:
extension: .Rmd
format_name: rmarkdown
format_version: '1.2'
jupytext_version: 1.11.2
kernelspec:
display_name: R
language: R
name: ir
editor_options:
chunk_output_type: console
---
# Lab: Regression Trees
## BART
The `tree` library is used to construct classification and regression trees.
```{r chunk1}
library(tree)
```
## Bayesian Additive Regression Trees
```{r chunk2}
library(ISLR2)
set.seed(1)
train <- sample(1:nrow(Boston), nrow(Boston) / 2)
```
```{r chunk3}
library(BART)
x <- Boston[, 1:12]
y <- Boston[, "medv"]
xtrain <- x[train, ]
ytrain <- y[train]
xtest <- x[-train, ]
ytest <- y[-train]
set.seed(1)
bartfit <- gbart(xtrain, ytrain, x.test = xtest)
```
Next we compute the test error.
```{r chunk4}
yhat.bart <- bartfit$yhat.test.mean
mean((ytest - yhat.bart)^2)
```
```{r chunk5}
ord <- order(bartfit$varcount.mean, decreasing = T)
bartfit$varcount.mean[ord]
```
Results
Run -> Run All gives
> mean((ytest - yhat.bart)^2)
[1] 21.73309
Knit gives
## [1] 15.94718
Tried "Clear Knit Cache", Knit on Save, always ## [1] 15.94718, run code in RStudio command by command. Always consistent [1] 21.73309
