End to end machine learning model deployment using flask

machine_learning_deployment_12.py

# Machine learning model development

# XGBoost model
xgb_model = xgb.XGBClassifier(
    objective = 'binary:logistic',
    use_label_encoder = False
)

# Define parameters range
params = {
    'eta': np.arange(0.1, 0.26, 0.05),
    'min_child_weight': np.arange(1, 5, 0.5).tolist(),
    'gamma': [5],
    'subsample': np.arange(0.5, 1.0, 0.11).tolist(),
    'colsample_bytree': np.arange(0.5, 1.0, 0.11).tolist()
}

# Make a scorer from a performance metric or loss function
scorers = {
    'f1_score': make_scorer(f1_score),
    'precision_score': make_scorer(precision_score),
    'recall_score': make_scorer(recall_score),
    'accuracy_score': make_scorer(accuracy_score)
}

# k-fold cross validation
skf = KFold(n_splits = 10, shuffle = True)

# Set up the grid search CV
grid = GridSearchCV(
    estimator = xgb_model,
    param_grid = params,
    scoring = scorers,
    n_jobs = -1,
    cv = skf.split(X_train, np.array(y_train)),
    refit = 'accuracy_score'
)

# Fit the model
grid.fit(X = X_train, y = y_train)

# Best parameters
grid.best_params_

# Create a prediction of training 
predicted = grid.predict(X_val)

# Model evaluation - training data
accuracy_baseline = accuracy_score(predicted, np.array(y_val))
recall_baseline = recall_score(predicted, np.array(y_val))
precision_baseline = precision_score(predicted, np.array(y_val))
f1_baseline = f1_score(predicted, np.array(y_val))
print('Accuracy for baseline   :{}'.format(round(accuracy_baseline, 5)))
print('Recall for baseline     :{}'.format(round(recall_baseline, 5)))
print('Precision for baseline  :{}'.format(round(precision_baseline, 5)))
print('F1 Score for baseline   :{}'.format(round(f1_baseline, 5)))