I’m confused. In the example, it said
Here it will be 3/2 = 1.5
Recall in this case, the number of items for which there is a prediction = 2/5 = 0.4
What is 1.5 here then?
2 Likes
If it’s kind of a validation set then why are there missing target values for some of the user_id in Training Data.csv. For eg., there’s no target label for user_id = 4 in Training Data Target.csv.
Am I missing something?
Hi @zer-0-ne,
Think of it like this. You have 100 user ids. You are given the correct answers of predicted categories of 60 ( the training data target). You have to predict for the rest, 40, on which you will be given a score.
I have taken indicative figures, but I hope you get the idea of what I am saying.
1 Like
I try to change my 2nd and 3rd recommendation of product keeping first same but I am getting same answer in each case. is this only depended on first product only.
No, it is not only dependent on the first product. You can see the scores of other participants on the leaderboard. heir MRR and precision is not the same. So I think it is quite unique for your approach.
Anyways precision and mrr being equal is an advantage. Right? You just need to increase precision your MRR will increase automatically.
Best of luck @amang251314
@nasir.afroze Out of all the submissions made by a participant for public leaderboard, which submission would be considered for private leaderboard (best public score, best private score, last submission or any other selection criteria)?
Is it Okay if i do not calculate MRR and Precision in the Notebook, I am not from a coding heavy background and I might depend on APIs , which may differ from library to library, is it okay if I submit only the Top3 predictions alongside my Training Notebook?
Hi @data007,
Thanks for reaching out.
Submission for the best score of the public leaderboard will be considered.
TIll now, I am not able to find how to evaluate the score on our end. We have only one true value which a user will buy but we have to predict three. we can find the score of the single true product as is this single product there in our prediction(Top-3) or not? but what about remaining two. we don’t have true value for the remaining two. how do we find that?
I hope, you are getting my point.
I have very naive question 
, I have created a recommendations system in which If I give a product name it will predict next 3 products , but I am not able to do the prediction considering user_id , it’s competition so I can’t explain much but how to create a CSV file which will add top3 prediction , I am getting prediction but not in CSV format . How to do it ?
Hi @amang251314,
You need to predict the correct product. If it comes in the first place ( it has your top predction) then no need to do anything further. But if it is at 3rd place then you need to tweak your model for getting it to 1st.
I hope this helps.
Hi @sarabhian6431 ,
I think you are facing an issue with the file format. You can easily convert your excel/other files into CSV with help of online tools.
I hope it helps,
Thanks for reaching out
- Am not able to understand for what will use that train target csv file.
- Taking what as our correct ans or label will train our model.
@nasir.afroze
Same here. Did you find any possible reasons leading to this behaviour, any suggestion would be helpful ?
What is the difference between public and private leaderboards??
1 Like
Pubic Leaderboard The Pubic leaderboard is live during the hackathon and displays your score based on the dataset provided. The private leaderboard will open after the end of the hackathon and will display your score based on a new dataset which we not provided you.
Hi @nasir.afroze,
Now that the competition has ended, can we get to know about the top submissions? Their approach and if possible the submitted notebooks?
Yes, We will organize a award ceremony and a debriefing session for this hackathon after completion of round 2. Then you can get access to all top submissions file or notebook.