Wow @tomwetherell you must have some special seed or mythical features. Congrats, once again you find a way to nail it. Except for some setback in May you have been consistently very high up on the LB. But if I remember correctly May played havoc with everybody.
Congrats that is really good.
@ihar you are also doing really well, congrats. Not to forget @tomy4reel and @JuliusFx - all of these players seem to be really consistent.
Well, the minis are wrapping up and the big one starts soon, any bets on who will take that cake?
Yes, Tom is doing really well, he has a secret sauce it seems :).. For me the volatility in some sub-categories seems to be my greatest challenge. June was transport, July was Housing and Utilities. I don't know how many more surprises awaits me..
I vaguely recall, some time ago, I looked at few time series techniques, just to get a benchmark, and on headline CPI the t+1 RMSE was ~ 0.5, so, at least on headline, I have it in my mind that you can squeeze out 0.5.
For the big comp there probably will be surprises for every model. Since you need to use the same model repeatedly it does average things out a bit over time if you have a good model, but, given the few forecast points, also not too much, so surprises will matter ...
@JuliusFx time to polish that fancy model of yours ... the big race is about to start.
I seem to have run out of ideas, but let see
Thanks a lot! Very happy with this month as I used my new model that I'd been working on for a couple of the categories.
Looking forward to the main compeition. I've still got a couple of ideas that I'm working on, but it's proving to be a challenge to beat the models I made in the first month. Best of luck to everyone for that!
Thanks - if only there was time to implement and test all the ideas ... my pipeline is a bit clunky and was implemented too hastily, so I think I am going to stop here.