QUESTION: Can someone explain utility of columns from model.predict() result?

[RamiroGhilino]

I'm working with prophet and i want to store in my DB the result of prediction some data, i just want to know if all the columns in the dataframe that returns from model.predict() have any particular use case or i can drop them. If someone could point any article or documentation where i can find the description of every column that would be great.

[AlexandroLuis]

Hello, the choice depends on your analysis needs, but in essence:

ds: This column depicts the dates for which you intend to make predictions. yhat: In this column, you'll find the forecasted values for the target variable. yhat_lower and yhat_upper: These columns furnish the lower and upper limits of the predicted values. Other columns house residuals, seasonal components, and additional data.

If your primary objective is displaying the predicted values alone, focusing on yhat should suffice.

[RamiroGhilino]

Thanks for the swift reply, but i was looking for the explanation of these "other columns", mainly "additive_terms" and "multiplicative_terms". I would also like to know if trends columns can be achived whit a SQL Sentence (something like a group by days for weekly trend or something like that). Thanks once again :)