Riiid! Answer Correctness Prediction Kaggle Challenge: 4th Place Solution Summary
This solution provides a competitive benchmark for the educational technology industry, specifically for predicting student answer correctness.
This paper presents a solution to the Riiid! Answer Correctness Prediction Kaggle challenge, achieving an AUC score of 0.817 and ranking 4th on the private leaderboard. The solution is a single transformer-based model incorporating time-aware attention, concatenated input sequence embeddings, and continuous feature embeddings.
This paper presents my solution to the challenge "Riiid! Answer Correctness Prediction" on Kaggle hosted by Riiid Labs (2020), which scores 0.817 (AUC) and ranks 4th on the final private leaderboard. It is a single transformer-based model heavily inspired from previous works such as SAKT, SAINT and SAINT+. Novel ingredients that I believed to have made a difference are the time-aware attention mechanism, the concatenation of the embeddings of the input sequences and the embedding of continuous features.