After applying LASSO regularization to select a linear regression model,

Which of the following is correct (only one answer)?

1. It is recommended to use the Ordinary Least Squares (OLS) method to estimate the regression coefficients for the model with the selected predictors.
2. It is recommended to make predictions using the model with the estimated regression coefficients obtained from LASSO and not from the Ordinary Least Squares (OLS) method.
3. The estimated regression coefficients from Lasso are more efficient than those provided by the Ordinary Least Squares (OLS) method.