Struct gbdt::decision_tree::Data

pub struct Data {
    pub feature: Vec<ValueType>,
    pub target: ValueType,
    pub weight: ValueType,
    pub label: ValueType,
    pub residual: ValueType,
    pub initial_guess: ValueType,
}

A training sample or a test sample. You can call new_training_data to generate a training sample, and call new_test_data to generate a test sample.

A training sample can be used as a test sample.

You can also directly generate a data with following guides:

1. When using the gbdt algorithm for training, you should set the values of feature, weight and label. If Config::initial_guess_enabled is true, you should set the value of initial_guess as well. Other fields can be arbitrary value.

2. When using the gbdt algorithm for inference, you should set the value of feature. Other fields can be arbitrary value.

3. When directly using the decision tree for training, only “SquaredError” is supported and you should set the values of feature, weight, label and target. label and target are equal. Other fields can be arbitrary value.

4. When directly using the decision tree for inference, only “SquaredError” is supported and you should set the values of feature.

Fields

feature: Vec<ValueType>

the vector of features

target: ValueType

the target value of the sample to be fit in one decistion tree. This value is calculated by gradient boost algorithm. If you want to use the decision tree with “SquaredError” directly, set this value with label value

weight: ValueType

sample’s weight. Used in training.

label: ValueType

sample’s label. Used in training. This value is the actual value of the training sample.

residual: ValueType

used by LAD loss. Calculated by gradient boost algorithm.

initial_guess: ValueType

used by gradient boost. Set this value if Config::initial_guess_enabled is true.

Implementations

impl Data

pub fn new_training_data(
    feature: Vec<ValueType>,
    weight: ValueType,
    label: ValueType,
    initial_guess: Option<ValueType>
) -> Self

Generate a training sample.

feature: the vector of features

weight: sample’s weight

label: sample’s label

initial_guess: initial prediction for the sample. This value is optional. Set this value if Config::initial_guess_enabled is true.

Example

use gbdt::decision_tree::Data;
let data1 = Data::new_training_data(vec![1.0, 2.0, 3.0],
                                1.0,
                                2.0,
                                Some(0.5));
let data2 = Data::new_training_data(vec![1.0, 2.0, 3.0],
                                1.0,
                                2.0,
                                None);

pub fn new_test_data(feature: Vec<ValueType>, label: Option<ValueType>) -> Self

Generate a test sample.

label: sample’s label. It’s optional.

Example

use gbdt::decision_tree::Data;
let data1 = Data::new_test_data(vec![1.0, 2.0, 3.0],
                                Some(0.5));
let data2 = Data::new_test_data(vec![1.0, 2.0, 3.0],
                                None);

Trait Implementations

impl Clone for Data

fn clone(&self) -> Data

Returns a copy of the value.

const fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Data

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Data

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>where
    __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Data

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>where
    __S: Serializer,

Serialize this value into the given Serde serializer.