Feedback / Suggestions / Experience on Online Hackathon 3.0


#1

Hope you all had a great time and experience competing for the top spot. We loved the experience of hosting this challenge and would want to refine the experience further based on your feedback and suggestions.

Please share your thoughts in general and around the questions mentioned below specifically and we will use them to go back to drawing board.

Overall Experience:

  1. How would you rate the overall experience? Would you recommend these hackathons to other fellow data scientists? If yes, how much would you score us on a scale of 10? If no, why?
  2. What do you think about the duration of the contest? Which kind of contest / hackathon would you prefer - week / month long contest with on-going engagement or a day / weekend long contest with high engagement?
  3. How difficult was the dataset and the problem? What is your total data science experience till date?

Features / changes:

Which of these features would you want to be included in upcoming hackathons:

  • Ability to form teams and participate in teams
  • Enable more than one solutions to be submitted
  • Ability to see performance of your model on higher percentage of test data for a penalty in score
    Any other features you would want?
    Also, please rank order them in order of their importance to you.

Looking forward to hear from you.

P.S. We will float a separate thread for approach to problem and learning from the competition post its completion.

Regards,
Kunal


#3

Overall Experience:

  1. Overall experience of the hackathon has been very good and would recommend to others. I would give a score of 7.5.
  2. The duration of the contest should depend on the problem statement, The current schedule of weekend long contest is fine. But longer time frame could be considered with a higher prize money.
  3. Weekend type problem requires a quick and dirty fix to the problem inspite the difficulty level. So it may not be possible to dig really deep into the dataset given the time constraints.

Features / changes:
Team formation should be allowed for longer time frame contests and not for such short duration ones.
More than one solution should be allowed on the lines of kaggle.
Verification of data on higher percentage is not reqd. 30% is more than enough.


#4

Feature changes? Weeklong hackathons. :frowning: Sadly, I wasn’t as free as I thought I would be, this weekend.


#5

#6

1). It was an awesome experience. I will surely recommend these hackathons to others. I will rate 8 out of 10.
2) I will prefer week/month long contest with on-going engagement. This gives flexibility to work on it as and when time permits.
3) I am a newbie in data science. Have completed some online courses. Overall the dataset looked simple. Except the Var columns which did not make much sense to me.

Features/Changes:
I would go for :

  1. Enable more than one solutions to be submitted
  2. Ability to see performance of your model on higher percentage of test data for a penalty in score

#7

Hi Kunal,
I think the registration for the competition should be open until the competition ends. It’ll give chances to the people who heard ‘late’ about the competition.
-Sachinmm


#8

First of all great work on organizing these hackathons! The competitive spirit is very high among the participants! However I had a few suggestions:

  1. We should be able to have a look at the history of our submissions and be able to write comments on it. This is far better than having to keep a record locally
  2. Also, why is there a need for all the participants to submit their code after the competition ends. Doesn’t make sense. In any case whenever the participants upload their csv file. The system must be ranking them on both the public as well as the private data set. So all you need to do is, inform the top 3 or top 5 competitors to submit their code. This will ensure that they have used only open-source software on the data/ or if you want to cross-validate their results.

#9

Great experience. The amount one learns in just a few days is incredible.

Personally, I really like the new two stage format. Its a great innovation. Its a combination of the speed hacking experience as well as the longer well thought out approach. Also, if one is busy on the weekend , one still gets a second chance to work on the problem.