  
# random-forests
A Rust library for [Random Forests](https://en.wikipedia.org/wiki/Random_forest).
Support for generic impurity measures:
[x] Entropy
[x] Gini
## installation
Add to your `Cargo.toml`:
```text
randomforests = "*"
```
## usage
Add the crate and `RandomForests` to your code:
```rust
extern crate randomforests;
use randomforests::RandomForest;
```
Create a `Dataset` as collection of `Item`s:
```rust
let mut dataset = Dataset::new();
let mut item1 = Item::new();
item1.insert("lang".to_string(), Value { data: "rust".to_string() });
item1.insert("typing".to_string(), Value { data: "static".to_string() });
dataset.push(item1);
let mut item2 = Item::new();
item2.insert("lang".to_string(), Value { data: "python".to_string() } );
item2.insert("typing".to_string(), Value { data: "dynamic".to_string() } );
dataset.push(item2);
let mut item3 = Item::new();
item3.insert("lang".to_string(), Value { data: "haskell".to_string() });
item3.insert("typing".to_string(), Value { data: "static".to_string() });
dataset.push(item3);
```
Initialise the classifier and train it classifier by passing the `Dataset`, a `TreeConfig`, the number of trees and the data subsample size:
```rust
let mut config = TreeConfig::new();
config.decision = "lang".to_string();
let forest = RandomForest::build("lang".to_string(), config, &dataset, 100, 3);
```
Create a question as an `Item`:
```rust
let mut question = Item::new();
question.insert("typing".to_string(), Value { data: "static".to_string() });
```
And get the predictions:
```rust
let answer = RandomForest::predict(forest, question);
// answer = {Value { data: haskell }: 48, Value { data: rust }: 52}
```