select all correct answers

In April 2021, Coinbase filed to go public on the Nasdaq under the symbol "COIN". The Coinbase data scientists wanted to predict their IPO's opening price from firm-specific and market-level features. They split the data frame into an 80% training set and 20% testing set. then hit three models of different sizes. The models' predictive performance (in terms of RMSE) is summarized in this table: The models' predictive performance (i Parameters In-sample RMSE Out-of-sample RMSE 5 $20.51 $23.38 14 $17.37 $18.10 3 47 $14.01 $26.57 Which of the following is true of this set of models? (1) Here we see evidence that simpler models generally show less decline in performance with the testing data relative to performance with training data. (2) Model 1 has the best in-sample RMSE. (3) The model with the best overall predictive performance is Model 2