Challenge 21 Megathread

For any and all questions relating to Challenge 21 :point_down: post away!

3 Likes

I was wondering if it is possible to bot in these competitions

more or less macro like autohotkey or some other sophisticated program

(post deleted by author)

With multiple choice, it seems to me like the best thing to do would be to remove the time stamp for the team prize and instead draw randomly from all those who tied. It’d help emphasize actually executing/learning the code and not needing to be already experienced with these concepts. At least for the coding challenge you had to have proper code to submit and pass all the tests, so the “brute force” option wouldn’t be quite as possible.

I certainly didn’t have any hope to win the team challenge as someone who is brand new to everything data.

3 Likes

Yeah, I’m still trying to figure it out. I have no idea what I’m looking at. It feels like the review (using everything we’ve learned so far) should’ve been the final challenge, not this. :woman_shrugging:

3 Likes

I like this approach, especially given different teams are in different time zones. Alternative would be to set it up similar to their Java Script challenge, where you have to submit code instead of multiple choice answers. Although it can still be “gamed” (one team member solves it and gives solution to others), at least that will encourage team spirit and collaboration (everyone joins a zoom session and try to code it out together).

1 Like

If you’re stuck on Q2, try looking here and searching for “evaluate”.

3 Likes

I agree a draw for the team prize should take place. I watched the submissions come in. My team finished in under a minute, but by then 2 other teams had already won.

While I agree I think that’s the fair thing to do, I don’t think LHL can change it due to their terms and conditions for the prizing. It’s something for them to consider at the next data challenge.

Maybe instead of one winning team getting 1600, could have multiple teams getting a smaller prize like 50 or 100 per member.

I am stuck with Q1 any hints much appreciated.

i am puting in the values for best_state, best_fitness = mlrose.genetic_alg(problem_fit, mutation_prob = 0.2,
max_attempts = 100, random_state = 2) and my alg gets stuck does not output anything.

I don’t think that line is supposed to output anything.

Try
print(best_state)
and see what happens. If you can print something, it’s working

Yes i tried the print statement doesn’t do anything ?

Its working now, seems like it takes time to compute.

sorry but this one I really did not understand. I think the problem should basically be solved with:

fitness_dists = mlrose.TravellingSales(distances = dist_list)
problem_fit = mlrose.TSPOpt(length = 16, fitness_fn = fitness_dists,
                            maximize=False)
best_state, best_fitness = mlrose.genetic_alg(problem_fit, random_state = 2, mutation_prob = 0.3, max_attempts = 100)

But I do not get anything similar to the alternatives. Instead, I get 130 hours and the

fitness_dists = mlrose.TravellingSales(distances = dist_list)
our_travel = np.array([0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16])
fitness_dists.evaluate(our_travel)

as 139.38

Have looked at the docu, and googled it, and still do not understand how they fixed the start city on the answers or the final stop. Quite frustrating, I must say.

Any hints of what must be actually done?

1 Like

You’re on the right path. I see a small error in your code that might be causing the difference in your result.

Are you sure about the length?

2 Likes