KNN, cross validation, grid search

What is cross validation (cross validation)
Cross validation ： The training data you will get , Divided into training and validation sets . The following is an example ： Divide the data into 4 Share , One of them is used as verification set . And then pass by 4 Time ( Group ) Test of , Change the validation set every time . Or get 4 The results of the group model , Take the average as the final result . also called 4 Crossover verification .

The training set is divided into verification set and training set
Fixed test set
Can't improve the accuracy of the model
Intelligence increases credibility

The grid search ------- Tuning parameters

Confirmation of hyperparametric problems, etc

Usually , There are many parameters that need to be specified manually （ Such as k- In the nearest neighbor algorithm K value ）, This is called superparameter . But the manual process is complicated , So we need to preset several super parameter combinations for the model . Each group of super parameters was evaluated by cross validation . Finally, the optimal combination of parameters is selected to establish the model .

API
sklearn.model_selection.GridSearchCV(estimator, param_grid=None,cv=None)
Detailed search for the specified parameter value of the estimator
estimator： Estimator objects
param_grid： Estimator parameters (dict){“n_neighbors”:[1,3,5]}
cv： Specify a few fold cross validation
fit： Input training data
score： Accuracy rate
Result analysis ：
bestscore__: The best results in cross validation
bestestimator： The best parametric model
cvresults: The accuracy results of verification set and training set after each cross validation

explain ：
For example, ten
Take nine at a time for training , Take one to verify
Each time take a different one as the verification set , You can do 10 Fold validation
Get ten different results , Average.

If it is n Share , You can do n Secondary cross validation

The grid search ： The optimal combination is obtained by cross validation

Here are some Inline code slice .

iris = load_iris()
# 2、 Basic data processing  --  Divide the data set 
x_train, x_test, y_train, y_test = train_test_split(iris.data, iris.target, random_state=22)
# 3、 Feature Engineering ： Standardization 
#  Instantiate a converter class 
transfer = StandardScaler()
#  call fit_transform
x_train = transfer.fit_transform(x_train)
x_test = transfer.transform(x_test)
# 4、KNN The predictor process 
#  4.1  Instantiate the predictor class 
estimator = KNeighborsClassifier()

# 4.2  Model selection and tuning —— Grid search and cross validation 
#  Prepare the hyper parameters to be adjusted 
param_dict = {"n_neighbors": [1, 3, 5]}
estimator = GridSearchCV(estimator, param_grid=param_dict, cv=3)
# 4.3 fit Data for training 
estimator.fit(x_train, y_train)
# 5、 Evaluate the effect of the model 
#  Method a： Compare the predicted results with the real values 
y_predict = estimator.predict(x_test)
print(" Compare the predicted results with the real values ：\n", y_predict == y_test)
#  Method b： Direct calculation accuracy 
score = estimator.score(x_test, y_test)
print(" Direct calculation accuracy ：\n", score)

print(" The best results in cross validation ：\n", estimator.best_score_)
print(" The best parametric model ：\n", estimator.best_estimator_)
print(" Accuracy results after each cross validation ：\n", estimator.cv_results_)