Summary
The chapter covers using grid search to systematically explore combinations of model hyperparameters with cross-validation in order to identify the best parameter settings.

General domain of usage
machine learning

Grid Search is a fundamental technique in the field of AutoML for systematically exploring the hyperparameter space of machine learning models.

Its primary purpose is to automate the process of finding the best combination of hyperparameters for a given model, which can significantly improve model performance.

In the context of AutoML, Grid Search helps you avoid manual trial-and-error by evaluating all possible parameter combinations within a specified grid, allowing for a more objective and reproducible approach to hyperparameter tuning.

from sklearn import datasets
from sklearn.model_selection import GridSearchCV
from sklearn.svm import SVC

# Load the iris dataset
iris = datasets.load_iris()
X = iris.data
y = iris.target

# Define the parameter grid for SVM
param_grid = {
    "C": [0.1, 1, 10],
    "kernel": ["linear", "rbf"],
    "gamma": [0.01, 0.1, 1]
}

# Create a GridSearchCV object with SVC
grid_search = GridSearchCV(SVC(), param_grid, cv=3)
grid_search.fit(X, y)

print("Best parameters found:", grid_search.best_params_)
print("Best cross-validation score:", grid_search.best_score_)

**Grid Search** evaluates all combinations of hyperparameters, which can become computationally expensive as the number of parameters and their possible values increases. It guarantees finding the best combination but requires exponentially more time with larger grids. While thorough, it may not be practical for large parameter spaces or limited resources. Alternatives like Random Search or Bayesian Optimization can be more efficient.

Note

import unittest
from sklearn.model_selection import GridSearchCV
from sklearn.neighbors import KNeighborsClassifier

def _dynamic_test(test_case, condition, success_msg, failure_msg):
    if condition:
        test_case._testMethodName = success_msg
        test_case.assertTrue(True, success_msg)
    else:
        test_case._testMethodName = failure_msg
        test_case.fail(failure_msg)

class TestGridSearch(unittest.TestCase):
    def setUp(self):
        try:
            from user_code import grid_search, best_params, best_score
            self.grid_search = grid_search
            self.best_params = best_params
            self.best_score = best_score
        except ImportError:
            self.fail("Variables 'grid_search', 'best_params', and 'best_score' must be defined.")

    def test_is_grid_search(self):
        _dynamic_test(
            self,
            isinstance(self.grid_search, GridSearchCV),
            "GridSearchCV object created successfully.",
            "grid_search must be an instance of GridSearchCV."
        )

    def test_estimator_type(self):
        _dynamic_test(
            self,
            isinstance(self.grid_search.estimator, KNeighborsClassifier),
            "KNeighborsClassifier used as estimator.",
            "grid_search must use KNeighborsClassifier."
        )

    def test_best_params_type(self):
        _dynamic_test(
            self,
            isinstance(self.best_params, dict),
            "Best parameters stored as dictionary.",
            "best_params must be a dictionary."
        )

    def test_best_score_value(self):
        _dynamic_test(
            self,
            isinstance(self.best_score, float) and self.best_score > 0.8,
            "Cross-validation score is valid and above 0.8.",
            "best_score must be a float greater than 0.8."
        )

test_main.py

Master the essentials of automated machine learning (AutoML) by exploring core concepts, optimization strategies, and hands-on frameworks. Learn to automate preprocessing, model selection, and hyperparameter tuning using leading Python tools.

Explore the core ideas behind AutoML, its workflow, and how it builds on traditional ML pipelines.

Dive into automated hyperparameter tuning methods and their impact on model performance.

Get hands-on with leading AutoML frameworks and see how they automate the ML pipeline.

Assess AutoML results, compare frameworks, and discuss deployment and limitations.

Grid Search Essentials

Oplossing