确定导致训练/测试数据集中得分最高的邻居数量

Question

enter image description here

I am working on GridSearchKNN case study and am confused on how to "Identify the number of neighbors that resulted in the max score in the training/testing dataset." This is my first time working with KNN.

I Tried:

from sklearn.metrics import accuracy_score
b_m = knn.fit(X_train, y_train)
y_pred = knn.predict(X_test)
print(accuracy_score(y_test, y_pred))

Answer 1

There are some options.

Option 1 - Using argmax .

best_k = np.argmax(test_scores) + 1

Option 2 - Update best k in iteration.

best_score = -1
for i in range(1, 10):
    knn = KneighborsClassifier(i)
    knn.fit(X_train, y_train)
    score = knn.score(X_test, y_test)
    if score > best_score:
        best_score = score
        best_k = i

If you want to find the best k only (that is, you don't want scores of not best k), the option 2 is better.