Machine Learning project to recommend the right mobile plan for Megaline customers: Smart or Ultra.
The model learns from customer behavior data to support plan upgrades and reduce legacy-plan usage.

Build a classification model to predict:

• is_ultra (1 = Ultra, 0 = Smart)

Minimum target:

• ✅ Accuracy ≥ 0.75

Dataset:

• users_behavior.csv

• Load and explore data
• Split into train/validation sets
• Train baseline models
• Tune hyperparameters
• Evaluate using accuracy (and review confusion matrix)

• Decision Tree
• Random Forest ✅ (final)
• Logistic Regression

Final model: RandomForestClassifier

• Validation Accuracy: 0.8085

This exceeds the required threshold and provides a solid balance of performance and stability.