Enigma Data Science Competition organized by Dept of Computer Science, IIT BHU as a part of Codefest'19.
I, in a team of 2, got first prize among first/second years and securing an overall rank of 18 among 500+ teams, which included industry professionals as well.
Train.csv- this had research papers with several attributes and their citation counts.
Author.csv- Each research paper had several authors and various attributes about these were given in this.
Test.csv- we had to predict citation counts for research papers with given attributes.
For a particular research paper, used author data to gather top 3 common interest fields, and then merged that with train data.
Used embedding layer to convert top 3 topics to embedded vector. Then we applied a simple neural model with activation function selu.
.