# Part $3$: Succcess Rate Function

In Part $4,$ you will be frequently asked to use simulation to estimate an agent's success rate under a given policy. To simplify this process, we will create a function to run such a simulation.

## $3.$A $-$ Define Function

Please define a function named $`$success$\_$rate$`$ with five parameters named $`$env$`,`$policy$`,`$episodes$`,`$max$\_$steps$`,$ and $`$random$\_$state$`.$ The function should perform the steps described below.

$1.$ Set the NumPy random seed to $`$random$\_$state$`.$

$2.$ Create a variable called $`$goals$`,$ setting it to $0.$

$3.$ Run a $`$for$`$ loop for a number of iterations indicated by the $`$episodes$`$ parameter. The loop should complete the following steps in each iteration:

$*$ Generate an episode for the environment instance $`$env$`,$ following the policy given by the $`$policy$`$ parameter. To avoid infinite loops, set $`$max$\_$steps$=$max$\_$steps$`.$

$*$ If the episode resulted in the agent finding the goal, increment $`$goals$`.$

$4.$ After the loop completes, calculate and return the observed success rate for the agent.

## $3.$B $-$ Test Function

Test your function by calling it on the $`$FrozenPlatform$`$ environment from Part $2$ along with the opimal policy found for that environment using value iteration. Use $10,000$ episodes, set $`$max$\_$steps$=200`,$ and set $`$random$\_$state$=1`.$

Print the message below with the blank filled in with the appropriate value, rounded to $4$ decimal places. If your function was implemented correctly, you should get a success rate of $0\mathrm{.}4114.$

When following the optimal policy, the agent's success rate was $\_\_\_\_.$