ml_knn

ml_knn(
    x,
    y,
    newdata,
    k,
    regression=True,
    what='prediction',
    seed=None,
    realizations=1000,
    alpha=0.1,
)

k-nearest-neighbors regression or classification.

Mirrors the C# KNearestNeighbors class: predicts from the k training rows closest to each new observation – an inverse-squared-distance weighted average of their responses for regression, or their most common response for classification.

Parameters

Name	Type	Description	Default
x	array_like	Training predictors, one row per observation.	required
y	array_like	The training response, one value per row of x.	required
newdata	array_like	Predictors to predict for, with the same number of columns as x.	required
k	int	The number of neighbors.	required
regression	bool	True for a weighted average; False for the modal class.	True
what	('prediction', 'neighbors', 'bootstrap', 'intervals')	Which result to return: the prediction (the default), the indices of each new observation’s neighbors, a prediction from one bootstrap resample of the training data, or bootstrapped prediction intervals.	"prediction"
seed	int	PRNG seed, used by "bootstrap" and "intervals"; None uses the clock.	None
realizations	int	The number of bootstrap resamples for "intervals".	1000
alpha	float	The interval level for "intervals": 0.1 gives a 90% interval.	0.1

Returns

Name	Type	Description
	numpy.ndarray	For "prediction" and "bootstrap", a 1-D array. For "neighbors", a 2-D array of 0-BASED training-row indices with one row per row of newdata and k columns. For "intervals", a 2-D array with four columns in the order lower, median, upper, mean – note "intervals" always reports percentiles, even for a classifier, because upstream has no classification branch there.

Examples

>>> from corehydropy import ml_knn
>>> x = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
>>> y = [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
>>> float(ml_knn(x, y, newdata=[4.5], k=2)[0])
45.0