Machine Learning Fundamentals

Data Preprocessing

• Handling Missing Values:
  • AI dataset: 30% missing values in categorical features, imputed using K-Nearest Neighbors (KNN) imputation
  • Titanic dataset: 20% missing values in age feature, imputed using median imputation
  • BostonHousing dataset: no missing values
• Feature Scaling:
  • AI dataset: standardized using StandardScaler
  • Titanic dataset: normalized using Min-Max Scaler
  • BostonHousing dataset: standardized using StandardScaler
• Feature Selection:
  • AI dataset: selected top 10 features using mutual information
  • Titanic dataset: selected top 5 features using recursive feature elimination
  • BostonHousing dataset: selected top 8 features using recursive feature elimination

Regression Models

• Simple Linear Regression:
  • AI dataset: R² = 0.75, mean squared error (MSE) = 10.5
  • Titanic dataset: R² = 0.42, MSE = 15.1
  • BostonHousing dataset: R² = 0.85, MSE = 5.2
• Multiple Linear Regression:
  • AI dataset: R² = 0.85, MSE = 9.2
  • Titanic dataset: R² = 0.50, MSE = 13.5
  • BostonHousing dataset: R² = 0.92, MSE = 4.5
• Ridge Regression:
  • AI dataset: alpha = 0.1, R² = 0.82, MSE = 10.2
  • Titanic dataset: alpha = 0.5, R² = 0.45, MSE = 14.2
  • BostonHousing dataset: alpha = 0.01, R² = 0.90, MSE = 4.8

Classification Algorithms

• Logistic Regression:
  • AI dataset: accuracy = 0.80, F1-score = 0.82
  • Titanic dataset: accuracy = 0.75, F1-score = 0.78
• Decision Trees:
  • AI dataset: accuracy = 0.85, F1-score = 0.88
  • Titanic dataset: accuracy = 0.80, F1-score = 0.82
• Random Forest:
  • AI dataset: accuracy = 0.90, F1-score = 0.92
  • Titanic dataset: accuracy = 0.85, F1-score = 0.88

Questions

• How do you handle missing values in a dataset?
• What is the purpose of feature scaling in machine learning?
• How do you select the most important features in a dataset?
• What is the difference between simple linear regression and multiple linear regression?
• How do you regularize a linear regression model?
• What is the difference between logistic regression and decision trees in classification problems?
• How do you evaluate the performance of a classification model?
• What is the role of hyperparameter tuning in machine learning?
• How do you handle class imbalance in a classification problem?
• What is the difference between precision and recall in classification evaluation metrics?