First, I’m not good at coding and English.(Sorry!, and this is my first question post.)
I wanted to make KNN Classifier in Python, google colab, not using Ai modules.
So I coded it for three hours fighting MANY Errors.
This is my code.

dlist={'index_id' :[],'distance':[]}
def find_near_target(df,k):
    #print('sort values and head cut_4.1')
    id_lst = df.sort_values('distance').head(k+1)[1:]['index_id']
    #print('#print id list',id_lst)
    #print('done_4.1')
    #print('make nlist_4.2')
    nlst=[]
    for i_d in id_lst.values:
        nlst.append(y_train.iloc[i_d])
    #print('#print n list',nlst)
    #print('done_4.2')
    #print('reshape_4.3')
    nplst = np.array(nlst).reshape(-1,1)
    #print('done_4.3')
    #print('return')
    return np.argmax(np.bincount(nplst[0]))

def knn_algorithm(feature,target):
    dlist={'index_id' :[],'distance':[]}
    #print('knn algoritm..._0')
    #print('destance calculate_1')
    for i in range(X_train.shape[0]//13):
        dlist['index_id'].append(i*13)
        dlist['distance'].append(math.sqrt(sum((X_train.iloc[i*13]-feature)**2)))
    #print('done_1')
    #print('reshape_2')
    #print('done_2')
    #print('reshape to dataframe_3')
    dflist = pd.DataFrame(dlist)
    #print('done_3')
    #print('find near target_4')
    pre_target = find_near_target(dflist,7)
    #print('done_4')
    #print('compare target_5')
    if (pre_target == target):
        #print('done_5')
        #print('true')
        return 0
    else :
        #print('done_5')
        #print('false')
        return 1

count=0
for x in range(X_test.shape[0]//27):
    count += knn_algorithm(X_test.iloc[x*27],y_test.iloc[x*27])

print(1-count/(X_test.shape[0]//27))

I used Mushroom Dataset (Binary Classification), in Kaggle.(https://www.kaggle.com/datasets/prishasawhney/mushroom-dataset)
I wonder why my code takes too much time to perform and why much less time in imported AI model.
I really want to know difference between mine and others.
Thankyou for reading, special thanks to respondent.

My handmade KNN code is too slow and I want to know why.