train_cols = data_adult.columns.difference([target])
model = KNeighborsClassifier()
params = {'n_neighbors': range(1,50)}
cv = StratifiedKFold(shuffle=True, n_splits=5)
GS_kNN = GridSearchCV(estimator=model, param_grid=params, cv=cv, scoring='roc_auc')
GS_kNN.fit(data_adult[train_cols], data_adult[target])
means = GS_kNN.cv_results_['mean_test_score']
stds = GS_kNN.cv_results_['std_test_score']
params = GS_kNN.best_estimator_
error = 1.95 * stds
plt.plot(params['n_neighbors'], means, color = 'blue')
plt.plot(params['n_neighbors'], means - error, color = 'green')
plt.plot(params['n_neighbors'], means + error, color = 'green')
plt.fill_between(params['n_neighbors'], means - error, means + error, color = 'green')
plt.xlabel('n_neighbors')
plt.ylabel('Average quality')
plt.title('Average quality for kNN model')
plt.show()