hard20 pts · first answer countsregressionmodel-interpretation

Coefficients that keep flipping

In a linear regression, "ad_spend_tv" and "ad_spend_total" are highly correlated. Coefficients change sign when you add or remove data, but predictions stay accurate. What is happening?

A.The model is overfitting so predictions must be wrongB.The data has too few rows to compute coefficients at allC.Multicollinearity: the model cannot separate the effect of correlated features, so individual coefficients are unstable even though combined predictions are fineD.Linear regression cannot handle money values

Discussion 0

Join free to post in the discussion.

Coefficients that keep flipping

Discussion 0

Related questions