Explain classification lifecycle and CTR modeling

You are interviewing for a Machine Learning Engineer role. Discuss the following machine-learning topics in a structured way: 1. Describe one practical implementation of a bag-of-words text feature pipeline. Include tokenization, vocabulary construction, handling rare or unseen words, sparse storage, and weighting choices such as raw counts or TF-IDF. 2. Explain out-of-bag (OOB) evaluation in ensemble methods such as bagging or random forests. How are OOB samples formed, and how can they be used for validation? 3. Suppose you need to build a binary classification model to predict click-through rate (CTR). Explain the full workflow from problem definition to deployment, including data collection, feature engineering, model selection, training, calibration, and evaluation. 4. More generally, if asked to build a classification model from scratch, walk through every major step and mention appropriate techniques or model choices at each stage. 5. If the model's online performance drops after deployment, how would you investigate and debug the issue? Cover model, data, serving, and product-level causes.