Create ML Model using Python

  • Status: Closed
  • Prize: $150
  • Entries Received: 10
  • Winner: yatishdua

Contest Brief

I need a Machine Learning model to classify the probability of success of the events Y in the spreadsheet attached based on the variables X for each event on the same spreadsheet.

Y is a random variable (Y=1 for success or Y=0 for failure) that is influenced by the X variables. Ideally each event should be classified in 1 of 7 classes, being the middle class 0 representing that the event has the same probability of the average of all events (around 36,5%). Higher class, (+1, +2, +3) should show an increased probability of success while lower classes( -1, -2, -3) should have a lower probability. (The classification model and/or number of classes is open for suggestion).

The best model will be chosen based on its capability to differentiate events with lower probabilities from the ones with higher probabilities of success. For example, class -3, will be better of the more events it has and the lower probability of success it has. Class +3 on the other side should also have the most events possible , but with the highest probability of success. Class 0 should have the fewest events and be around the dataset average.

I also attached a spreadsheet with an example of the desired outcome. The example should not be used as a target, but only as a reference of the desired output. I believe the outputs of the example could be improved.

Make sure the model is not overfitting the data. The models will be tested against another set of the same data that is not included in the spreadsheet.

I require the model to be implemented in python. Tensorflow is optional, but preferred.

Recommended Skills

Winner
Featured Contest
Post a Contest Like This

Employer Feedback

“He delivery the solution as requested.”

Profile image bddmendonca, Brazil.

Top entries from this contest

View More Entries

Public Clarification Board

  • meghraj9
    meghraj9
    • 3 years ago

    Please have a look at #18 and #19

    • 3 years ago
  • yatishdua
    yatishdua
    • 3 years ago

    Please have a look upon #17 Entry.

    • 3 years ago
  • yatishdua
    yatishdua
    • 3 years ago

    Please have a look upon #14 Entry.

    Thanks for the explanation , it really cleared lots of doubts.

    • 3 years ago
  • bddmendonca
    Contest Holder
    • 3 years ago

    If anyone requires, you can contact me through the chat

    • 3 years ago
    1. vijaykrishna0497
      vijaykrishna0497
      • 3 years ago

      Hi sir, can we chat. Please message me. I will try my best.

      • 3 years ago
  • msavinash1139
    msavinash1139
    • 3 years ago

    Hi, a small clarification. What does the random variable REPRESENT? For example, what does the class +2(or -2) represent? I understood it holds some percentage of y values, but what does it mean/represent? And How does +3 differ from +1?

    • 3 years ago
  • yatishdua
    yatishdua
    • 3 years ago

    There is a lot of ambiguity in the problem statement. It will be really better if you could explain it more or give an example.

    • 3 years ago
    1. bddmendonca
      Contest Holder
      • 3 years ago

      Hi, what are your questions exactly? I believe the description of the contest has all the info you need, I also uploaded a spreadsheet with an example of the desired outcome.

      Basically what I need is a model that given a new set of events it could predict the probability of success of the new events with a fairly degree of accuracy. Simply overfitting the current dataset will not solve the issue.

      • 3 years ago
  • yatishdua
    yatishdua
    • 3 years ago

    Please have a look upon #6 Entry.

    • 3 years ago
    1. bddmendonca
      Contest Holder
      • 3 years ago

      Hi Yatish, i looked entry #6. I don't know what strategy did you use to classify the events, but I am sure they are overfitting the dataset. The Y variable is a random variable, there is no way you could perfectly know with 100% certainty the outcome of the 40k events a priori like your model did.

      • 3 years ago
  • msavinash1139
    msavinash1139
    • 3 years ago

    Hi, I didn't understand your idea of 7 classes. I believe this is a binary classification problem, meaning there are only 2 classes. Could you elaborate the idea of 7 classes here?

    • 3 years ago
    1. vijaykrishna0497
      vijaykrishna0497
      • 3 years ago

      Yes, please this is prediction of a binary variable and we also need to classify binary random variable into seven clusters. Is it correct ?

      • 3 years ago
    2. bddmendonca
      Contest Holder
      • 3 years ago

      Hi Avinash, sorry I didn't see your question before. It is not exactly a binary classification problem, the Y variable is a random variable. My objective is to separate each event into a class based on the probability of the outcome. Each class will have a different probability of success.

      • 3 years ago
  • vijaykrishna0497
    vijaykrishna0497
    • 3 years ago

    please check #11 .

    • 3 years ago
  • vijaykrishna0497
    vijaykrishna0497
    • 3 years ago

    please check my entries #8 , #9 , #10 . give any feedback, I will improve if anything.

    • 3 years ago
  • vijaykrishna0497
    vijaykrishna0497
    • 3 years ago

    please check #8

    • 3 years ago
  • ShahrearBinAmin
    ShahrearBinAmin
    • 3 years ago

    Same question as Avinash?

    • 3 years ago

Show more comments

How to get started with contests

  • Post your contest

    Post Your Contest Quick and easy

  • Get tons of entries

    Get Tons of Entries From around the world

  • Award the best entry

    Award the best entry Download the files - Easy!

Post a Contest Now or Join us Today!