import pyutil as pyu

pyu.get_local_pyinfo()

'conda env: ml312-2024; pyv: 3.12.7 | packaged by Anaconda, Inc. | (main, Oct  4 2024, 13:27:36) [GCC 11.2.0]'

# !pip install deap

!pip show deap | grep "Version:"

Version: 1.4.1

import pandas as pd
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split, cross_val_score
from sklearn.metrics import accuracy_score
from deap import base, creator, tools, algorithms

# Load the Titanic dataset
url = 'https://raw.githubusercontent.com/datasciencedojo/datasets/master/titanic.csv'
data = pd.read_csv(url)

# Preprocessing
# Select features and target
features = ['Pclass', 'Sex', 'Age', 'SibSp', 'Parch', 'Fare', 'Embarked']
X = data[features]
y = data['Survived']

# Handle missing values
X.loc[:, 'Age'] = X['Age'].fillna(X['Age'].median())
X.loc[:, 'Embarked'] = X['Embarked'].fillna(X['Embarked'].mode()[0])

# Convert categorical variables to numerical
X = pd.get_dummies(X, columns=['Sex', 'Embarked'], drop_first=True)

# Train-test split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Define the Genetic Algorithm
# Create the fitness function
def evaluate(individual):
    n_estimators, max_depth, min_samples_split = individual

    # Ensure parameters are valid
    n_estimators = max(1, int(n_estimators))  # Minimum 1 tree
    max_depth = max(1, int(max_depth))  # Minimum depth of 1
    min_samples_split = max(2, int(min_samples_split))  # Minimum split of 2

    model = RandomForestClassifier(
        n_estimators=n_estimators,
        max_depth=max_depth,
        min_samples_split=min_samples_split,
        random_state=42
    )
    # Use 5-fold cross-validation
    scores = cross_val_score(model, X_train, y_train, cv=5, scoring='accuracy', n_jobs=-1)
    return np.mean(scores),

# Create the Genetic Algorithm structure
creator.create("FitnessMax", base.Fitness, weights=(1.0,))  # Maximize accuracy
creator.create("Individual", list, fitness=creator.FitnessMax)

/home/rajaraman/miniconda3/envs/ml312-2024/lib/python3.12/site-packages/deap/creator.py:185: RuntimeWarning: A class named 'FitnessMax' has already been created and it will be overwritten. Consider deleting previous creation of that class or rename it.
  warnings.warn("A class named '{0}' has already been created and it "
/home/rajaraman/miniconda3/envs/ml312-2024/lib/python3.12/site-packages/deap/creator.py:185: RuntimeWarning: A class named 'Individual' has already been created and it will be overwritten. Consider deleting previous creation of that class or rename it.
  warnings.warn("A class named '{0}' has already been created and it "

toolbox = base.Toolbox()

# toolbox.register("attr_int", np.random.randint, 100, 500)  # n_estimators range
# toolbox.register("attr_depth", np.random.randint, 5, 20)   # max_depth range
# toolbox.register("attr_split", np.random.randint, 2, 10)   # min_samples_split range

# Register attributes for hyperparameters
toolbox.register("attr_int", np.random.randint, 100, 500)  # n_estimators range
toolbox.register("attr_depth", np.random.randint, 5, 20)   # max_depth range
toolbox.register("attr_split", np.random.randint, 2, 10)   # min_samples_split range

# Register individual and population
toolbox.register("individual", tools.initCycle, creator.Individual,
                 (toolbox.attr_int, toolbox.attr_depth, toolbox.attr_split))
toolbox.register("population", tools.initRepeat, list, toolbox.individual)

# Register genetic operators
toolbox.register("mate", tools.cxBlend, alpha=0.5)  # Crossover
toolbox.register("mutate", tools.mutGaussian, mu=0, sigma=10, indpb=0.2)  # Mutation
toolbox.register("select", tools.selTournament, tournsize=3)  # Selection
toolbox.register("evaluate", evaluate)  # Evaluation function


# toolbox.register("individual", tools.initCycle, creator.Individual,
#                  (toolbox.attr_int, toolbox.attr_depth, toolbox.attr_split))
# toolbox.register("population", tools.initRepeat, list, toolbox.individual)

# toolbox.register("mate", tools.cxBlend, alpha=0.5)          # Crossover
# toolbox.register("mutate", tools.mutGaussian, mu=0, sigma=10, indpb=0.2)  # Mutation
# toolbox.register("select", tools.selTournament, tournsize=3)  # Selection
# toolbox.register("evaluate", evaluate)

# Genetic Algorithm execution
population = toolbox.population(n=20)  # Initial population of 20
NGEN = 10  # Number of generations
CXPB, MUTPB = 0.5, 0.2  # Crossover and mutation probabilities

for gen in range(NGEN):
    offspring = algorithms.varAnd(population, toolbox, cxpb=CXPB, mutpb=MUTPB)
    fits = list(map(toolbox.evaluate, offspring))
    for fit, ind in zip(fits, offspring):
        ind.fitness.values = fit
    population = toolbox.select(offspring, k=len(population))
    print(f"Generation {gen + 1}: Best Fitness = {max(ind.fitness.values for ind in population)}")

Generation 1: Best Fitness = (0.8300305328474342,)
Generation 2: Best Fitness = (0.832837584950261,)
Generation 3: Best Fitness = (0.832837584950261,)
Generation 4: Best Fitness = (0.832837584950261,)
Generation 5: Best Fitness = (0.832837584950261,)
Generation 6: Best Fitness = (0.832837584950261,)
Generation 7: Best Fitness = (0.8356446370530879,)
Generation 8: Best Fitness = (0.8356446370530879,)
Generation 9: Best Fitness = (0.8356446370530879,)
Generation 10: Best Fitness = (0.8356446370530879,)

# Get the best individual
best_individual = tools.selBest(population, k=1)[0]
print("Best Hyperparameters:", {
    "n_estimators": int(best_individual[0]),
    "max_depth": int(best_individual[1]),
    "min_samples_split": int(best_individual[2]),
})

Best Hyperparameters: {'n_estimators': 540, 'max_depth': 6, 'min_samples_split': 12}

# Train the model with the best hyperparameters
best_model = RandomForestClassifier(
    n_estimators=int(best_individual[0]),
    max_depth=int(best_individual[1]),
    min_samples_split=int(best_individual[2]),
    random_state=42
)
best_model.fit(X_train, y_train)

# Evaluate on the test set
y_pred = best_model.predict(X_test)
print("Test Accuracy:", accuracy_score(y_test, y_pred))

Test Accuracy: 0.8100558659217877