Hi all, We've come up with a simple imputation approach that we think has advantages in some cases. Overall, it performs decently (see NittanyLions entries on live leaderboard) and outperforms Avocado on several datasets. We were planning to use our predictor alongside the provided Avocado imputations as input into a learning method. However, I'm not sure that we have enough time left to complete and optimize that approach. So - if anyone already has an ensemble approach in place where you're using Avocado and other inputs, and if you'd like to test our imputed tracks as an additional predictor, please get in touch with me. My email is mahony@psu.edu. Our only condition is that if our tracks help your method to increase in performance, you will provide us with an opportunity to merge teams (as allowed by the challenge rules). Shaun