1. Design a Machine Learning Task for Trello
You are given a scenario where users have two types of search queries in Trello: passive (such as reading and browsing) and active (such as modifying). How would you implement a machine learning task from scratch to classify these queries? Discuss what data you would use and why, how to convert data from strings to numerical data, how to combine binary and numerical features, how to handle the situation if there are no labels, how to deal with imbalanced data, which model to use (mentioning logistic regression as a baseline and then building a neural network), the differences between logistic regression and neural network, and which evaluation metric to use, including the definitions of precision and recall.
