# Import libraries and filter warnings
import warnings
warnings.filterwarnings('ignore')

# Data setup
import pandas as pd
url = 'https://raw.githubusercontent.com/datasciencedojo/datasets/master/titanic.csv'
titanic = pd.read_csv(url)
print('Shape:', titanic.shape)
titanic.head()

# See the target distribution
titanic['Survived'].value_counts().plot(kind='bar', title='Survival Counts')

# Prepare features and target
predictors = ['Pclass', 'Age', 'SibSp', 'Fare']
target = 'Survived'
X = titanic[predictors]
y = titanic[target]
X = X.fillna(X.mean())
X.head()

# Split data for training and testing
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

# Simple decision tree, no tuning yet
from sklearn.tree import DecisionTreeClassifier
tree = DecisionTreeClassifier(random_state=42)
tree.fit(X_train, y_train)
score = tree.score(X_test, y_test)
print('Test accuracy:', score)

# Import GridSearchCV
from sklearn.model_selection import GridSearchCV

# Set search parameter grid
param_grid = {
    'max_depth': [2, 4, 6, 8],
    'min_samples_split': [2, 5, 10],
    'criterion': ['gini', 'entropy']
}

# Set up the grid search
grid = GridSearchCV(
    DecisionTreeClassifier(random_state=42),
    param_grid,
    cv=5,
    scoring='accuracy'
)

# Fit the grid search
grid.fit(X_train, y_train)

# See the best parameters found
print('Best parameters:', grid.best_params_)
print('Best cross-val accuracy:', grid.best_score_)

# Test the best model on the test set
best = grid.best_estimator_
test_score = best.score(X_test, y_test)
print('Final test accuracy:', test_score)

# View all tested settings and results
results = pd.DataFrame(grid.cv_results_)
results[['params', 'mean_test_score', 'rank_test_score']].sort_values(by='mean_test_score', ascending=False).head()

# Use input() for simple user story
depth = input('Pick a tree depth to try (eg. 2, 3, 4): ')
custom_tree = DecisionTreeClassifier(max_depth=int(depth), random_state=42)
custom_tree.fit(X_train, y_train)
print('Test accuracy:', custom_tree.score(X_test, y_test))

# Tips: Avoiding errors if you make a typo
try:
    depth = int(input('Pick another tree depth: '))
except ValueError:
    print('That is not a valid number! Using depth=2.')
    depth = 2
tree2 = DecisionTreeClassifier(max_depth=depth, random_state=42)
tree2.fit(X_train, y_train)
print('Accuracy:', tree2.score(X_test, y_test))

# Best practices: Use less data or fewer options for speed
## Try a small grid for quick learning
quick_grid = {
    'max_depth': [2, 3],
    'min_samples_split': [2, 5]
}
quick_search = GridSearchCV(DecisionTreeClassifier(random_state=42), quick_grid, cv=3)
quick_search.fit(X_train, y_train)
print('Best found:', quick_search.best_params_)

# Troubleshooting: What if all results look the same?
if results['mean_test_score'].nunique() == 1:
    print('All parameter combos have the same accuracy! Try tweaking grid or preprocessing.')
    

# Extra: Use GridSearchCV with another model
from sklearn.ensemble import RandomForestClassifier
forest_grid = {
    'n_estimators': [10, 20],
    'max_depth': [2, 5]
}
forest_search = GridSearchCV(RandomForestClassifier(random_state=42), forest_grid, cv=3)
forest_search.fit(X_train, y_train)
print('Random forest best:', forest_search.best_params_)