How to get precision and recall from gridsearch results?

Question

So I ran the following code to tune a model (XGBoost). All I am looking for is how to calculate the precision and recall for the best model below.

How exactly do I implement the recall_score and precision_score from best model below? Do I need anything else besides what I have?

from sklearn.model_selection import PredefinedSplit
import numpy as np

# Combine the feature matrix in the training and validation data
X_train_valid = np.vstack((X_train, X_valid))

# Combine the target vector in the training and validation data
y_train_valid = np.append(y_train, y_valid)

# Get the indices of training and validation data
train_valid_idxs = np.append(np.full(X_train.shape[0], -1), np.full(X_valid.shape[0], 0))

# The PredefinedSplit
ps = PredefinedSplit(train_valid_idxs)


from xgboost.sklearn import XGBClassifier
models = {'xgbc': XGBClassifier(seed=42)}



###Pipeline Dictionary
from sklearn.pipeline import Pipeline
from sklearn.model_selection import GridSearchCV

pipes = {}

for acronym, model in models.items():
    pipes[acronym] = Pipeline([('model', model)])
    
    
    
param_grids = {}


##XGBoost
# The grids for eta LEARNING RATE
eta_grids = [.1]

# The grids for gamma
gamma_grids = [0.5]

# The grids for lambda
lambda_grids = [10 ** i for i in range(-1, 2)]

# Update param_grids
param_grids['xgbc'] = [{'model__eta': eta_grids,
                        'model__gamma': gamma_grids,
                        'model__lambda': lambda_grids}]

# The list of [best_score_, best_params_, best_estimator_] obtained by GridSearchCV
best_score_param_estimator_gs = []

for acronym in pipes.keys():
    # GridSearchCV
    gs = GridSearchCV(estimator=pipes[acronym],
                      param_grid=param_grids[acronym],
                      scoring='roc_auc',
                      n_jobs=-1,
                      cv=ps,
                      return_train_score=True)
        
    # Fit the pipeline
    gs = gs.fit(X_train_valid, y_train_valid)
    
    # Update best_score_param_estimator_gs
    best_score_param_estimator_gs.append([gs.best_score_, gs.best_params_, gs.best_estimator_])

Given this best performing model, all I am seeking is how I would now calculate precision and accuracy on this best model.

If it makes it any easier, let's say that the best predictor was:

• ETA: .1
• GAMMA: .5
• LAMBDA: 10