Making a better NN

Error in the in$-$class NN using qsec, hp is about $2\mathrm{.}38$

CART tree analysis identifies the $3$ most important predictors of

am as wt$,$ disp, mpg

$*$

Re$-$do the NN using wt$+$disp$+$mpg as predictors

$-$Use set.seed$(5)$ before neuralnet for reproducibility

What is the new error using wt$+$disp$+$mpg as predictors?

$-$Please give answer to $2$ decimals, like $1\mathrm{.}23$

Now build a NN using all variables as predictors $($am $.)$

$-$Use set.seed$(5)$ before neuralnet for reproducibility

Use nn$net.result to view activation levels for $am=1.$

What is the activation level for the Volvo$142$E$?$

$-$Please give answer to $2$ decimals, like $1\mathrm{.}23$