Chicago Machine Learning Study Group Message Board › 2011-01-12 Meeting Notes & my M4 results
We spent the meeting talking through the M4 project. I proposed a very simple first step which I walked the group through and performed (live!).
That method was to simply take the average by time of day for each segment for all of the data in the more recent data set and make predictions based on that data.
I did this and submitted it tonight. My RMSE was 246.626. This put this submission at 167th out of the 214 teams submitting. There was one other entry with precisely the same RMSE. I'm betting pretty strongly that he used the same technique.
Thoughts on improving this?