We spent this week trying to tweak our programs in various ways to get better results. On method one, we tried changing the weight given to the votes from guessed unknown. We tried 0.7/DSD and 0.9/DSD. The former gave us an average of 48.86% from fourteen runs and the latter gave us an average of  49.13% from nine runs. We also tried changing the number of runs for some of our programs, but that did not change the accuracy much. Another thing we tried doing was keeping all the initial predictions from the first iteration, and using all of them in our final votes (we previously voted once, and used the one vote for further predictions). We are still in the midst of trying new methods and collecting more data.