A retail company's daily sales forecasting model has been producing inaccurate forecasts for three weeks. An AWS Glue job appends actual sales and model predictions to S3 each day. The ML team is investigating model degradation in a SageMaker Studio notebook. Which visualization will most accurately show model degradation over time?

A line chart showing the model's weekly Mean Absolute Error (MAE) is the most effective visualization for identifying model degradation over time. MAE is a common metric for regression tasks like sales forecasting, and plotting it over time directly reveals trends in prediction accuracy. An increasing MAE would clearly indicate degradation. Histograms of daily sales (option 1) would show the distribution of sales but wouldn't directly quantify prediction error or its trend. Histograms of model errors (option 2) could show a shift in error distribution but are less intuitive for tracking degradation over time compared to a single metric like MAE. Scatter plots of daily sales versus model error (option 4) might reveal patterns in errors but are not ideal for visualizing a trend in overall model performance over an extended period.