Negative coef_lower and coef_upper values for regressor in Prophet

I am using the Prophet package in R to forecast retail sales using a regressor called Spend, which represents the spend/revenue ratio. I have added the Spend regressor to the model using the add_regressor() function with mode = "additive" and standardize = TRUE.

However, I am getting negative values for the coef_lower and coef_upper parameters of the Spend regressor, which indicates that the model is estimating a negative effect of Spend on retail sales. This is not consistent with my domain knowledge and the observed relationship between Spend and retail sales.

I have tried adjusting the prior.scale parameter of the add_regressor() function to place a stronger prior belief on the Spend coefficient being positive, but this has not resolved the issue.

Is there a way to prevent negative coef_lower and coef_upper values for the Spend regressor in Prophet, or another way to represent Spend so that the model can take it into consideration?

Start with the add_regressor() documentation and the coefficient output containing coef_lower and coef_upper. Investigate whether Prophet documents a supported positivity constraint or an alternative regressor representation; done means the supported behavior and limitation are clearly established with a focused example.

Written by the indexing model from the issue text.