When developing a linear regression model in BigQuery ML to predict a customer's likelihood of purchasing your company's products, city names are a key predictive component. However, the data must be organized into columns for both training and serving the model. What is the most efficient method to prepare this data?