Use the validation dataset to tune better hyperparameter for rbf $($gamma$).$ Use $C=10$ as before.

gammas $=$ # Add gammas to your choice

accs$=[]$

for gamma in gammas:

# Implement here

plt$.$ figure $($figsize $=(15,6)$

plt$.$plot$($gammas$,$ accs, color$=$'red'$)$

plt$.$xlabel$(\text{'}$gamma$\text{'})$

plt$.$ylabel$(\text{'}$accuracy$\text{'})$

plt$.$title$(\text{'}$Tuning $\text{'})$

plt$.$xticks $($gammas$)$

plt$.$ show $()$

Questions:

Without coding, only from the plot, will the accuracy change using the best $$ you found? Why?

Why did we use $C=10?$