Random forest in depth in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

How does a random forest reduce overfitting compared to a single decision tree?

Random forests build many decision trees and combine their results. Why does this help reduce overfitting compared to using just one tree?

A Because averaging many trees reduces variance and prevents the model from fitting noise in the training data.

B Because random forests use deeper trees that memorize the training data better.

C Because random forests only use a single feature for all splits, making the model simpler.

D Because random forests remove all noisy data points before training.

💡 Hint

Think about how averaging multiple guesses can make the final guess more stable.

Random forests reduce overfitting by averaging many trees trained on different random subsets of data and features. This averaging lowers variance and prevents the model from fitting noise.

❓ Predict Output

intermediate

Output of feature importance extraction from a random forest

What is the output of this code snippet that trains a random forest and prints feature importances?

ML Python

from sklearn.datasets import load_iris
from sklearn.ensemble import RandomForestClassifier

iris = load_iris()
X, y = iris.data, iris.target
model = RandomForestClassifier(n_estimators=10, random_state=42)
model.fit(X, y)
importances = model.feature_importances_
print([round(i, 2) for i in importances])

A [0.5, 0.2, 0.2, 0.1]

B [0.25, 0.25, 0.25, 0.25]

C [0.11, 0.03, 0.44, 0.42]

D [0.0, 0.0, 0.0, 1.0]

💡 Hint

Feature importances sum to 1 and reflect how useful each feature is for splitting.

The printed list shows the importance of each feature rounded to two decimals. The last two features have higher importance in this trained model.

❓ Hyperparameter

advanced

Which hyperparameter controls the number of features considered at each split in a random forest?

When training a random forest, which hyperparameter decides how many features are randomly selected to consider for splitting at each node?

A max_depth

B max_features

C min_samples_split

D n_estimators

💡 Hint

This parameter controls randomness in feature selection per split.

max_features controls how many features are randomly chosen at each split, adding randomness and diversity to trees.

❓ Metrics

advanced

Interpreting out-of-bag (OOB) error in random forests

What does the out-of-bag (OOB) error estimate in a random forest represent?

A The error on data not used to train each tree, providing an unbiased estimate of test error.

B The error on the training data used to build each tree.

C The error on a separate validation dataset provided by the user.

D The error after pruning the trees to reduce complexity.

💡 Hint

OOB samples are those left out when bootstrapping data for each tree.

OOB error is calculated using samples not included in the bootstrap sample for each tree, giving an unbiased estimate of model performance without needing a separate validation set.

🔧 Debug

expert

Why does this random forest model raise a ValueError?

Consider this code snippet:

from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier(n_estimators=100, max_features='auto')
model.fit(X_train, y_train)

Why does this code raise a ValueError?

A Because RandomForestClassifier requires max_depth to be set explicitly.

B Because n_estimators must be less than 50.

C Because X_train and y_train are not defined.

D Because 'auto' is not a valid value for max_features in RandomForestClassifier in recent sklearn versions.

💡 Hint

Check the allowed values for max_features in sklearn 1.2+.

In recent sklearn versions, 'auto' is deprecated for max_features in RandomForestClassifier. Use 'sqrt' instead.