What metrics would you use to evaluate the performance of a regression model and why?

During my final year project in college, I developed a regression model to predict student grades based on study hours and attendance. To evaluate the model's performance, I used Mean Absolute Error (MAE) and R-squared metrics. MAE helped me understand the average error in predictions, which was crucial for assessing real-world applicability since I wanted to ensure that students could rely on my predictions. R-squared provided insights into how well my model explained the variability in student grades. This experience taught me the importance of not just developing a model, but also effectively communicating its performance to my peers and professors.

In my part-time role as a data assistant at a retail store, I had the opportunity to analyze sales data using a regression model. I focused on metrics like Mean Squared Error (MSE) and R-squared to evaluate the model's accuracy in forecasting sales trends. Using MSE allowed me to understand how much my predictions deviated from actual sales, which was particularly important for inventory management. R-squared helped my team grasp how much of the sales variance was explained by factors like promotions and seasonality. This practical experience emphasized the need to choose the right metrics to inform business decisions effectively.

In my first job as a junior data analyst, I worked on a project to predict customer churn using regression analysis. We employed several metrics, including MAE and R-squared, to evaluate our model’s performance. MAE provided a straightforward interpretation of our prediction errors, which was essential for our presentations to stakeholders. R-squared allowed us to communicate how well our model fit the data, which was crucial for demonstrating the project's value. This experience reinforced my understanding of the importance of selecting appropriate metrics based on the business context and the specific questions we aimed to answer.