Build a machine learning notebook that predicts whether a user session will lead to a purchase. The training and test CSV files contain page counts, time spent, exit and dropoff rates, engagement score, promotion indicator, encoded user device and traffic fields, holiday flag, and the target `outcome`. The deliverables are a documented notebook and `submissions.csv` with `id` and binary `outcome` predictions. The evaluation metric is weighted F1. ```hint Hint 1 Start by naming the core entities, constraints, and success criteria. ``` ```hint Hint 2 Make the trade-offs explicit before going deep on implementation details. ``` ### Constraints & Assumptions - The target is binary. - The test set has no target labels. - Some fields may contain bugs or unwanted values. - Columns with more than 50 percent missing values or constant values should be dropped. - The notebook may use one or more models. ### Clarifying Questions to Ask - What counts as an unwanted value for each field? - Is class imbalance severe? - Should categorical integer fields be treated as categories or ordinal numbers? - Is model interpretability required? - What validation split should match the hidden evaluation? ### What a Strong Answer Covers ```premium-lock What a Strong Answer Covers ``` ### Follow-up Questions - How would you handle class imbalance? - How would you avoid leakage during preprocessing? - What baseline model would you start with? - How would you debug a low F1 score?