I am using Prophet for sales forecasting, but I am encountering an overfitting issue. My current model parameters are leading to an R-squared (R2) value of 1.00, indicating perfect fit, which suggests overfitting. When I try to reduce the changepoint_prior_scale (CPS) and fourier_order values to mitigate this, my error metrics (MAE, MSE, RMSE) increase significantly, leading to very large errors.