Boosted Trees in R

# Boosted Trees in R
### Dr. D’Agostino McGowan

---

<div class="my-footer">
<span>
Dr. Lucy D'Agostino McGowan
</span>
</div>

---

## Boosted trees in R

```r
library(tidymodels)
```

```r
*boost_spec <- boost_tree(
  mode = "classification", 
  tree_depth = 1, 
  trees = 1000, 
  learn_rate = 0.001, 
) %>% 
* set_engine("xgboost")
```

---

## Boosted trees in R

```r
library(tidymodels)
```

```r
boost_spec <- boost_tree( 
* mode = "classification",
  tree_depth = 1, 
  trees = 1000, 
  learn_rate = 0.001, 
) %>% 
  set_engine("xgboost") 
```

* Set the `mode` as you would with a bagged tree or random forest

---

## Boosted trees in R

```r
library(tidymodels)
```

```r
boost_spec <- boost_tree( 
  mode = "classification", 
* tree_depth = 1,
  trees = 1000,
  learn_rate = 0.001,
) %>% 
  set_engine("xgboost") 
```

* `tree_depth` here is the depth of each tree, let's set that to 1

---

## Boosted trees in R

```r
library(tidymodels)
```

```r
boost_spec <- boost_tree( 
  mode = "classification", 
  tree_depth = 1, 
* trees = 1000,
  learn_rate = 0.001,
) %>% 
  set_engine("xgboost") 
```

* `trees` is the number of trees that are fit, this is equivalent to `B`

---

## Boosted trees in R

```r
library(tidymodels)
```

```r
boost_spec <- boost_tree( 
  mode = "classification", 
  tree_depth = 1, 
  trees = 1000, 
* learn_rate = 0.001,
) %>% 
  set_engine("xgboost") 
```

* `learn_rate` is `\(\lambda\)`

---

## Fit the model

```r
model <- fit(boost_spec,
             HD ~ Age + Sex + ChestPain + RestBP + Chol + Fbs + 
               RestECG + MaxHR + ExAng + Oldpeak + Slope + Ca + Thal,               
             data = heart)
```

---

## <i class="fas fa-laptop"></i> `Boosting`

How would this code change if I wanted to tune `B` the number of bootstrapped training samples?

```r
boost_spec <- boost_tree( 
  mode = "classification", 
  tree_depth = 1, 
  trees = 1000, 
  learn_rate = 0.001, 
) %>% 
  set_engine("xgboost") 
```