The first algorithm for today's study--k-nearest neighbor algorithm

Def classify0 (inx,data_set,labels,k):

DATA_SET_SIZE=DATA_SET.SHAPE[0]

. shape[0] indicates How many rows to get Data_set,. shape[1] that gets the number of columns

also note The use of brackets after shape

Diff_mat=tile (InX, (data_size,1))-data_size

Tile can extend an array, the above code is the InX in the array copy data_size , anyway, it is not good to express, you can Baidu. Using this will save you from using a for loop.

Sq_diff_mat=dif_mat**2

Sq_distances=sq_diff_mat.sum (Axis=1)

Axis=1 means Adding the elements in each line

Axis=0 means Adding the elements in each column (default)

distance=sq_distance**0.5

Sorted_dist_indicies=distance.argsort ()

Sort the distance from small to large, but store the number in sorted_dist_indicies

class_count={}

Create a dictionary that stores The number of K predictor tags

For I in range (k):

Vote_label=labels[sorted_dist_indicies[i]]

Class_count[vote_label]=class_count.get (vote_label,0) +1

The. Get method returns the value of the key vote_label ,0 If the key is not in the dictionary , Add the key and assign a value of 0.

Sorted_class_count=sorted (Class_count.items (), Key=operator.itemgetter (1), reverse=true)

I use the python3, so the items used here , if it is python2, it will be used . Iteritems (), Key=operator.itemgetter (1), which represents the sort value ,reverse=true indicates a descending arrangement, and the above code returns an element as a tuple ( tuple) the list (list)

Return Sorted_class_count[0][0]

returns the label of the forecast

The first algorithm for today's study--k-nearest neighbor algorithm