In this section, you will train a linear model for a binary text classification task. Given an email, the model must determine whether it is a **phishing email (label = 1)** or a **legitimate email (label = 0)**.

Because the dataset is imbalanced (with legitimate emails appearing more frequently, as in real-world scenarios) the model’s performance will be evaluated using the F1-score, which balances precision and recall for the phishing class.

**Goal:** Your objective is to build and optimize a linear model that achieves the best F1-score. You may tune hyperparameters, apply preprocessing techniques, and experiment with different vectorization methods to improve overall performance.