Engineer and Impute ZIP Features

You are building a predictive model for a product team. For some users, you have address fields such as street, city, state, and ZIP code. 1. What features would you derive directly from the address, and what external public datasets could you join on ZIP code or geography to create additional model features? 2. If ZIP code is missing for some users, how would you handle those cases? In your answer, discuss: - useful geographic and socioeconomic features - high-cardinality encoding choices - when to drop vs. impute missing ZIP codes - missingness as a potentially informative signal - fairness, privacy, and leakage risks - how you would evaluate whether these features improve model performance