Mojo struct

KNN

@memory_only
struct KNN[metric: String = "euc"]

Classifier implementing the k-nearest neighbors vote.

Aliases

• MODEL_ID = 4
• metric_ids = List(String("euc"), String("man"), __list_literal__=NoneType(None))

Parameters

• metric (String): Metric to use for distance computation: Euclidean -> 'euc'; Manhattan -> 'man'.

Fields

• k (Int): Number of neighbors to use.
• search_depth (Int): Current KDTree implementation applies some approximation to its search results. Increasing search_depth can lead to more accurate results at the cost of performance.
• kdtree (KDTree[True, metric=metric])
• y_train (Matrix)

Implemented traits

AnyType, CV, Copyable, ImplicitlyDestructible, Movable

Methods

__init__

fn __init__(out self, k: Int = 3, search_depth: Int = 1)

Args:

• k (Int)
• search_depth (Int)
• self (Self)

Returns:

Self

Raises:

fn __init__(out self, params: Dict[String, String])

Args:

• params (Dict[String, String])
• self (Self)

Returns:

Self

Raises:

fit

fn fit(mut self, X: Matrix, y: Matrix)

Fit the k-nearest neighbors classifier from the training dataset.

Args:

• self (Self)
• X (Matrix)
• y (Matrix)

Raises:

predict

fn predict(mut self, X: Matrix) -> Matrix

Predict the class indices for the provided data.

Args:

• self (Self)
• X (Matrix)

Returns:

Matrix: Class indices for each data sample.

Raises:

save

fn save(self, path: String)

Save model data necessary for prediction to the specified path.

Args:

• self (Self)
• path (String)

Raises:

load

@staticmethod
fn load[type: UInt8](path: String) -> KNN[(load_from_mem KNN[metric].metric_ids[type])]

Load a saved model from the specified path for prediction.

Parameters:

• type (UInt8)

Args:

• path (String)

Returns:

KNN[(load_from_mem KNN[metric].metric_ids[type])]

Raises: